Search

The existing pig counting system based on image processing are seriously affected by light conditions, and the counting accuracy is poor when pigs are crowded and obscured.To realize the intelligent pig counting, the paper proposed a counting scheme based on the improved instance segmentation algorithm was proposed in this paper. Aiming at the problems of image illumination and target edge blur, Laplace operator was used to preprocess the images. The feature extraction network of MASK R-CNN network was improved by using Resnet-152 as the Mask R-CNN feature extraction network, and the original Feature Pyramid Network(FPN) was followed by a bottom-up enhancement path, which directly fused the low-level edge features with the high-level features to improve the recognition ability of the target edge bluring. The non maximum suppression process and loss function were optimized and improved to improve the segmentation accuracy. The experiments were carried out at three different real pig farms to verify the counting accuracy, respectively in Hebei province, Jilin province and Inner Mongolia. The size of the cage in Hebei pig farm was 5.5 m × 1.8 m, with an average of 12 pigs in a single pen, with a feeding density of 1.21 pigs/m²; the size of the cage in Jilin pig farm was 5.5 m × 3.9 m, with an average of 22 pigs in a single pen, and the average rearing density was 1.03 pigs/m²; the size of the cage in Inner Mongolia pig farm was 11.4 m × 5.28 m, with an average of 80 pigs in a single pen, the average feeding density was 1.32 pigs/m². The RGB camera is was installed on the top of the pen and acquired the image in daytime. 2 400 images were collected in total, and 2000 images were selected after image preprocessing, and 1 250 images of three pig farms were selected as the original data set according to the ratio of 2:2:1. The training set and verification set were enhanced to 1 500 and 150 images and 250 images for the test set, The experimental results showed that 98 images could realize exact counting and 2 images missed 1 pig in the Jilin pig farm, the accuracy of pig counting was 98%. In Hebei pig farm, 99 images could realize the exact counting and the accuracy of pig counting was 99%. For Inner Mongolia pig farms with high feeding density, the accuracy of pig counting was 86%, among the 50 test images, 7 images missed detection, 4 images missed 1 target, 3 images missed 2 targets, The results can provide the application of the artificial intelligent in agriculture field. [ABSTRACT FROM AUTHOR]

Depth estimation is critical to 3D reconstruction and object location in intelligent agricultural machinery vision system, and a common method in it is stereo matching. Traditional stereo matching method used low-quality image extracted manually. Because the color and texture in the image of field plant is nonuniform, the artificial features in the image are poorly distinguishable and mismatching could occur as a result. This would compromise the accuracy of the depth of the map. While the supervised learning-based convolution neural network (CNN) is able to estimate the depth of each pixel in plant image directly, it is expensive to annotate the depth data. In this paper, we present a depth estimation model based on the self-supervised learning to phenotype tomato canopy. The tasks of the depth estimation method were to reconstruct the image. The dense disparity maps were estimated indirectly using the rectified stereo pair of images as the network input, from which a bilinear interpolation was used to sample the input images to reconstruct the warping images. We developed three channel wise group convolutional (CWGC) modules, including the dimension invariable convolution module, the down-sampling convolution module and the up-sampling convolution module, and used them to construct the convolutional auto-encoder - a key infrastructure in the depth estimation method. Considering the shortage of manual features for comparing image similarity, we used the loss in image convolutional feature similarity as one objective of the network training. A CWGC-based CNN classification network (CWGCNet) was developed to extract the low-level features automatically. In addition to the loss in image convolutional feature similarity, we also considered the whole training loss, which include the image appearance matching loss, disparity smoothness loss and left-right disparity consistency loss. A stereo pair of images of tomato was sampled using a binocular camera in a greenhouse. After epipolar rectification, the pair of images was constructed for training and testing of the depth estimation model. Using the Microsoft Cognitive Toolkit (CNTK), the CWGCNet and the depth estimation network of the tomato images were calculated using Python. Both training and testing experiments were conducted in a computer with a Tesla K40c GPU (graphics processing unit). The results showed that the shallow convolutional layer of the CWGCNet successfully extracted the low-level multiformity image features to calculate the loss in image convolutional feature similarity. The convolutional auto-encoder developed in this paper was able to significantly improve the disparity map estimated by the depth estimation model. The loss function in image convolutional feature similarity had a remarkable effect on accuracy of the image depth. The accuracy of the disparity map estimated by the model increased with the number of convolution modules for calculating the loss in convolutional feature similarity. When sampled within 9.0 m, the root means square error (RMSE) and the mean absolute error (MAE) of the corner distance estimated by the model were less than 2.5 cm and 1.8 cm, respectively, while when sampled within 3.0m, the associated errors were less than 0.7cm and 0.5cm, respectively. The coefficient of determination (R²) of the proposed model was 0.8081, and the test speed was 28 fps (frames per second). Compared with the existing models, the proposed model reduced the RMSE and MAE by 33.1% and 35.6% respectively, while increased calculation speed by 52.2%. [ABSTRACT FROM AUTHOR]

The symptoms of the same tomato disease in different stages are obviously different, and different diseases show some similarities. Traditional pattern recognition methods can not reflect the dynamic changes of the pathological characterization, and the practicability is poor. To solve this problem, this paper proposed a Multi-Scale AlexNet recognition model for mobile platform based on convolutional neural network (CNN), and implemented a tomato leaf disease image recognition system for agricultural workers based on Android. Many parameters and large memory utilization of traditional AlexNet model are unfit for mobile platform, this paper adjusted the network structure of the traditional model by removing the local response normalization(LRN) layer, modifying the full connection layer, setting up different convolution kernel extraction features, designed a multi- scale recognition model based on the AlexNet. The model consists of 6 layers. It can optimize the training time and memory utilization and achieve high precision. After removing the LRN layer, there was a 30% decrease in running time. Extending the single convolution kernel into multi-scale (1 × 1,3 × 3,5 × 5,7 × 7) convolution kernel then fused at the first layer, removing full connection layer 6 and 7, and taking the place of global average pooling layer, then the model size was only 30.2 M. The forward propagation rate (FRP) and backward propagation rate (BPR) were reduced, and the global average pooling is better than the global maximum pooling on recognition accuracy. So the Multi-Scale AlexNet model used global average pooling in the 5th layer. In image preprocessing phase, in order to avoiding over fitting of the trained model caused by the unbalanced distribution of sample numbers, we had zoomed, flipped, color jittering, add noise and rotated the original pictures of dataset randomly to get the augmented dataset, and used 70% of the pictures as the train dataset and the rest as the validation dataset(20%) and test dataset(10%). These pictures were quantized to 224 × 224 dpi for Multi-Scale AlexNet training, and the original dataset and augmented dataset were used to train models. In order to validate the performance of the proposed model, comparative tests were done between Multi-Scale AlexNet and traditional pattern recognition method. It repeated 600 tests. The results showed that the CNN model achieved 92.7%, the high average recognition accuracy of each disease and each disease in the early, middle and late stages. Compared with the other CNN Net model(MobileNet, SequeezeNet, LeNet-5), the Multi-Scale AlexNet achieved the highest recognition accuracy, and reached 95.8% on the late stage disease dataset, but the SequeezeNet model used less memory. The MobileNet and SequeezeNet model reached lower recognition accuracy on the middle and late stages dataset, that because their convolution size was small. The recognition system was implemented on Android platform, and then test was done on field dataset. The results showed that the average recognition accuracy was 89.2%, its less value was due to the complex background of image. Then the system can meet the requirements of disease image recognition on mobile platform in production practice. The research results provide a method for disease image recognition based on convolution neural network, and provide a reference for automatic identification of crop diseases and engineering applications. [ABSTRACT FROM AUTHOR]

It is the key to prevent and control dairy cow diseases by real-time monitoring and detection of abnormal behavior early. In this field, it is the most common for the detection and recognition of lame behavior especially. However, the technology of the livestock abnormal behavior recognition is still facing the problem of poor real-time performance. Regarding the above problems, 2 improvement strategies were proposed in this paper. Firstly, a sparse overcomplete dictionary learning algorithm based on conjugate gradient pursuit-KSVD (CGP-KSVD) algorithm was proposed which was used to description and representation of lame behavioral semantic features. The proposed idea was to optimize search direction in the stage of sparse coding construction though introducing a conjugate gradient pursuit algorithm. Meanwhile, the proposed method could avoid the computational load caused by storing and calculating large-scale Hessen matrix effectively. By this way, it achieved the high convergence speed of the sparse overcomplete dictionary learning. Secondly, the basic features of representing the videos were extracted by the fusion of spatio-temporal interest points and spatio-temporal interest points extracted in the dense trajectory map. With the proposed method, it is possible to reduce the redundancy features and the computational load while retaining the rich details. To verify the proposed algorithm, we designed 3 experiments. Experiment 1 was used to achieve the recognition accuracy by comparing the proposed algorithm with the classical sparse dictionary learning algorithms. Experiment 2 and Experiment 3 were used to study the efficiency of the algorithm. In the experiments, 500 black and white adult cows were selected randomly and a total of 1200 video samples were achieved for training and testing by the manual marks. In Experiment 1, the average accuracy of the algorithm proposed in this paper is 5.5 percentage points higher than that of LLC (locality-constrained linear coding) and 3.9 percentage points higher than that of K-SVD. The fusion feature proposed in this paper was used as the basic features of LLC and K-SVD and the above methods were recorded as LLC + fusion feature and K-SVD + fusion feature. The experiment results show that the average accuracy of the proposed algorithm is 3.8 and 1.4 percentage points higher than that of LLC + fusion feature and K-SVD + fusion feature, respectively. In Experiment 2 and 3, the theoretical analysis results show that the algorithm proposed in this paper has the lowest computational time complexity, and the average response time of the CGP-KSVD algorithm is 0.043 s, faster than the 2 other algorithms, BP-KSVD (basis pursuit-KSVD) and OMP-KSVD (orthogonal matching pursuit-KSVD). And the 90-hour video test results show that the CGP-KSVD algorithm has the highest recognition rates of 93.3% and 92.7% respectively with the playback video and online test video, which are increased by 33.3 and 35.6 percentage points compared with the BP-KSVD algorithm, and 13.3 and 14.1 percentage points compared with the OMP-KSVD algorithm. [ABSTRACT FROM AUTHOR]

Ochotona curzoniae is an endemic species and key species in the alpine meadow of the Tibetan Plateau and also it is a main kind of organism that destroys the grassland ecology of the plateau. In order to prevent the dangers of the ochotona curzoniae, we need to study the living habits of ochotona curzoniae and investigate the degree of harm of ochotona curzoniae, and then we can control the number of ochotona curzoniae through the effective preventive measures. With the development of sensing technology and image processing, we can provide an objective basis through intelligent monitoring system to control the damage of ochotona curzoniae. The object detection of ochotona curzoniae is a key technology in the intelligent monitoring equipment because it can provide the object contour feature for behavior analysis of ochotona curzoniae. The object detection of ochotona curzoniae is very difficult, because the ochotona curzoniae video possesses the characteristics of complex background, low contrast, the object color with intensity inhomogeneity, diversity and mutability. The traditional object detection method cannot extract the object contours accurately. This paper presents a fast object detection method based on space-time domain. Firstly, the centroid position of the object in the current frame image is determined by the background subtraction, and then the rough segmented image and the initial contour are obtained based on the centroid position. The rough segmented image is segmented by the improved Chan-Vese model, and then we can obtain the object contours. In view of the fact that the level set function needs to be initialized in the process of improved Chan-Vese model, and the initialized computation is enhanced with the increase of the image scale, the centroid of the object is taken as the center to intercept the image containing the object as the roughly segmented image. Then, the improved Chan-Vese model is used to segment the roughly segmented image, so as to reduce the time consumption of Chan-Vese model segmentation. In addition, as Chan-Vese model can’t fully segment the image of object whose color is diverse and mutable, we use the improved Chan-Vese model to segment the roughly segmented image. The internal pixels of image evolution contours were processed by K-means clustering, and the clustering center point values were obtained. The internal fitting values of Chan-Vese model were constructed by the clustering center point values and the image mean filtered intensity information, thereby improving the adaptability of Chan-Vese model for complex object image segmentation．In addition, rectangular Dirac function was used to replace regularized Dirac function in the energy function of Chan-Vese model, and the calculation of level set evolution equation could be limited to the zero level set so as to avoid the influence of the image background disturbance on the segmentation result. In this paper, the video processing with 50 frames of images shows that the time consumption of this method is only 15.25 s, the average value of Dice similarity coefficient is 0.852 929, and the average value of Jaccard index is 0.744 57. In summary, the object detection method proposed in this paper can accurately extract the object contour and has a high real-time performance. [ABSTRACT FROM AUTHOR]

White Agaricus bisporus is an excellent source of the B vitamins, riboflavin, niacin, and pantothenic acid and also a good source of the dietary mineral phosphorus. It is one of the most commonly and widely consumed mushrooms in the world. The production of white Agaricus bisporus has been industrialized in China. However, during the last production chain, it needs a lot of labors to sort and grade the white Agaricus bisporus. The manual sorting and grading mode has many disadvantages such as larger error, low productivity, non-uniform standard, and so on. With the development of machine vision technology, it has been successfully used for automatic inspection and sorting, especially in agricultural industry due to its nondestructive characteristic. An automatic sorting and grading system based on machine vision was designed in this paper. Firstly, an automatic sorting and grading hardware system was designed. It included conveying mechanism, image acquiring system, control module and actuator. The conveying mechanism consists of the fixed support, conveyor, roller, driving unit, tension unit, cleaner and guide plate. The image acquiring system consists of a line scan camera, lens, light source and its controller. The control module consists of photoelectric sensor controlling the camera, electromagnetic valve and relay starter controlling the actuator. The actuator consists of the air compressor, duplex pieces, muffler, air cylinder, joints, gas channel, guide rod, slider, connector, flap, baffle, and so on. Secondly, it put forward an image algorithm based on the watershed method, Canny operator, OR operation and closed operation to determine the diameter of white Agaricus bisporus. The first watershed algorithm combining the global threshold segmentation method and maximum entropy threshold segmentation method is used to remove the shadow of image. The second watershed algorithm based on Canny operator, OR operator and closed operation is used to remove the disturbance of petiole. The minimum enclosing rectangle method is used to get the diameter of white button mushroom. Thirdly, a precise control strategy based on the conveyor speed, distance between trigger and flap piece, trigger time and algorithm processing time was designed. Finally, the software based on OpenCV 2.4.10 and Visual Studio 2010 was developed in this paper to acquire, analyze and process the image and output the control instruction to control the valve by USB-4761 module. It also has the visual parameter configuration function for camera, control module and grading standard. When the whole system starts at the first time, the light source of image acquiring system is adjusted by the light controller to make the image clear and stable. The air compressor is powered on and pressurized to the rated pressure of 0.3 MPa. The motor of conveyor is started and the speed of the conveyor belt is adjusted according to the actual production requirement. The white Agaricus bisporus goes into the guide bar with the conveyor belt. When the white Agaricus bisporus goes into the region of image acquiring, it triggers the industrial camera to scan. The image data are transmitted to the industrial computer by image capture card. The software analyzes the image on line. The analyzed result is sent to the related solenoid valve through the digital control module. When the white Agaricus bisporus arrives near the container of related grade, it is sorted into the container by the related flap piece driven by the solenoid valve. In order to validate the applicability and reliability, the test of grading performance and effect was carried out with the prototype of sorting and grading system at 12.7 m/min conveying speed and 1900 Hz line frequency. The results showed that the average maximum grading speed was 102.41 pieces/min, the accuracy of grading was 97.42%, the damage rate was 0.05% and the undetected rate was 0.96%. The grading speed improved by 38. 86%, the accuracy improved by 6.84% and the damage rate reduced by 0.13% compared to the manual grading. The system can stably and continuously operate. For long time grading, the advantage of intelligent system is more obvious due to the fatigue of labor. The whole system realizes the on-line automatically sorting and grading for fresh white Agaricus bisporus with the minimum destruction. [ABSTRACT FROM AUTHOR]

Accurate crop-load estimation is very important for efficient management of nutrients and harvest operations. Current machine vision techniques for crop-load estimation have achieved only limited success mostly due to partial occlusion, shape irregularity, varying illumination and multiple sizes. Detecting immature green fruit is a more challenging task for similar color of fruit and background. The key starting point of this paper for detecting immature citrus fruit was the observation that the light distribution on citrus fruit follows a general pattern in which the light intensity decreases with the distance from a local maximum due to specular reflection. Immature citrus fruit detection was achieved by detecting this pattern with concentric circles or parts of circles. This pattern was proposed with the maximally stable extremal region (MSER) method and validated by hierarchical contour analysis (HCA), and was the first proposed in this paper. The images were captured by a color camera under low natural light conditions with a flashlight, and the green component of the color images was used for further analysis. After smoothing the whole image by Weiner filter, the regions of interest (ROIs) in the image were extracted by the method of MSER. The ROIs detected by MSER were those whose support was nearly the same over a range of thresholds, so the regions on citrus fruit were detected by MSER for the pattern that the light intensity decreases stably and gently with the distance from a local maximum. However, many regions on leaves and background were also detected as ROIs and should be excluded in the next step. A novel algorithmic technique was proposed to remove these regions on background, and this method was named as the HCA. Firstly, shape analysis was used for each ROI and only those ROIs were considered as valid if the shape was nearly circular. Secondly, multiple levels of contours around each valid ROI were extracted and fitted with the circular Hough transform (CHT). Lastly, multiple fitted circles would be merged into one if their most parts were overlapped together, this step was called circle merging and the merged circles were considered as the last detected citrus fruits. The algorithm was tested on a testing dataset with 20 images and achieved the recall rate of 81.2% and the precision rate of 83.5%. The harmonic average of them was 82.3%. The result showed that 24.8 fruits were correctly detected, 5.7 fruits were not detected and 4.9 ROIs were wrong detected as fruit in each image on average. The processing time of the proposed method was 3.70 s totally on each image, on average, in which 0.57 s was used for MSER detection and 3.13 s was used for HCA. The result showed that the proposed method can detect green citrus fruit in a very difficult and challenging scene with so many fruits in one image and extensive partial occlusion. The good performance of partial occlusion tolerance of the proposed method in this paper is mainly due to that the proposed HCA doesn't use the shape of outer contour of fruit, but uses multiple concentric contours which come from the pattern of light intensity distribution on fruit surface. The research framework in this paper can give a novel thought on other green fruit detection besides citrus fruit. [ABSTRACT FROM AUTHOR]

The existing pig counting system based on image processing are seriously affected by light conditions, and the counting accuracy is poor when pigs are crowded and obscured.To realize the intelligent pig counting, the paper proposed a counting scheme based on the improved instance segmentation algorithm was proposed in this study. Aiming at the problems of image illumination and target edge blur, Laplace operator was used to preprocess the images. The feature extraction network of MASK R-CNN network was improved by using Resnet-152 as the Mask R-CNN feature extraction network, and the original Feature Pyramid Network(FPN) was followed by a bottom-up enhancement path, which directly fused the low-level edge features with the high-level features to improve the recognition ability of the target edge bluring. The non maximum suppression process and loss function were optimized and improved to improve the segmentation accuracy. The experiments were carried out at three different real pig farms to verify the counting accuracy, respectively in Hebei province, Jilin province and Inner Mongolia. The size of the cage in Hebei pig farm was 5.5 m × 1.8 m, with an average of 12 pigs in a single pen, with a feeding density of 1.21 pigs/m²; the size of the cage in Jilin pig farm was 5.5 m × 3.9 m, with an average of 22 pigs in a single pen, and the average rearing density was 1.03 pigs/m²; the size of the cage in Inner Mongolia pig farm was 11.4 m × 5.28 m, with an average of 80 pigs in a single pen, the average feeding density was 1.32 pigs/m². The RGB camera is was installed on the top of the pen and acquired the image in daytime. 2 400 images were collected in total, and 2000 images were selected after image preprocessing, and 1 250 images of three pig farms were selected as the original data set according to the ratio of 2:2:1. The training set and verification set were enhanced to 1 500 and 150 images and 250 images for the test set, The experimental results showed that 98 images could realize exact counting and 2 images missed 1 pig in the Jilin pig farm, the accuracy of pig counting was 98%. In Hebei pig farm, 99 images could realize the exact counting and the accuracy of pig counting was 99%. For Inner Mongolia pig farms with high feeding density, the accuracy of pig counting was 86%, among the 50 test images, 7 images missed detection, 4 images missed 1 target, 3 images missed 2 targets, The results can provide the application of the artificial intelligent in agriculture field. [ABSTRACT FROM AUTHOR]

Harris algorithm is a classical corner detection algorithm. It can extract corners of image quickly and has a certain degree of anti-noise ability, but it has corner location error to some extent. It needs to artificially set 2 threshold parameters, and it can not easily eliminate false corners such as edge points, so it has somewhat lower accuracy of corner detection. For above-mentioned reasons, a modified Harris corner detection algorithm based on auto-correlation matrix of image pixel was proposed in this paper, and the purpose was not only to solve the problem of the variability and randomness of setting thresholds for corner response function (CRF) and non-maximum suppression in Harris algorithm, but also to improve the accuracy of corner location. In our paper, the most important innovation is embodied in 2 aspects: One is avoiding to set 2 thresholds of traditional Harris corner detection algorithm artificially, the other is locating corner more accurately by modified non-maximum suppression method. Firstly, original image was filtered by directional filtering and Gaussian low-pass filtering, and feature corner image was constructed by calculating determinant of every pixel's auto-correlation matrix. Potential corners of image could be heightened effectively, which had more significant intensity than other surrounding pixels, and could be recognized easily in feature corner image. Secondly, in order to improve intelligent level of the modified algorithm, we selected adaptive OTSU algorithm to determine segmentation threshold. The segmentation threshold of feature corner image could be calculated by OTSU algorithm, and the pre-selected regions were obtained. So the search range of corner detection was significantly decreased. On the basis, an optimized non-maximum suppression method was adopted in our research, which could divide each pre-selected region into several 3×3 square subranges, and correct corners were extracted from potential corners of each square subrange, false corners were eliminated effectively. Finally, in order to validate the efficiency and reliability of the modified algorithm, 5 groups of comparison experiments were performed in our research. Five images, including generic image format (jpg, bmp), and multi-band remote sensing image format (GF-2 data), were selected to test performance of the modified algorithm and Harris algorithm, which contained the total of detection corners, the number of correct corners, the number of false corners, the number of omissive corners, and the detection rate of correct corners. According 5 groups of comparison experiments, the accuracy of corner detection in different types of images is improved, for crop vegetation remote sensing image, the accuracy of corner detection can reach 27.06 percentage points. We can draw a conclusion that the improved algorithm can not only calculate the optimal threshold automatically, but also locate the corners more accurately. Therefore, our modified algorithm can greatly improve the precision of corner detection. The proposed algorithm is more accurate and efficient than traditional algorithm, its adaptive characteristic makes it easy to be integrated in an image processing system or image registration module, and it has higher feasibility and application value. Experiments also show that there is some insufficiency to be improved in our research, for example, some corners in picture of cubes could not be detected correctly with either our modified algorithm or Harris algorithm. In our future research, we propose to partition an image into several sub-image blocks, and independently determine each sub-image block's segmentation threshold by OTSU algorithm, so that the corners not prominent in full image can be significantly strengthened in sub-image blocks, and can be detected correctly. The research could provide reference for agricultural remote sensing image data detection. [ABSTRACT FROM AUTHOR]

The breeding environment plays an important role in healthy growth and development of the pigs, and it is also a guarantee of the excellent traits. The growth of pig is the integrated result of their own traits and external environmental factors. For different environments, pigs will show different behaviors. It is found that the appropriate growth environment can be reflected by their degree of dispersion in the pens. When the environment is suitable, the pig group will lie down together. When the temperature is low, they will huddle together; and when the environmental temperature is high, the pigs will scatter around. Image segmentation is the important link to analyze pig behavior status based on computer vision technology, and pig image segmentation is an image representation that provides easy-to-understand and analysis for the behavior feature extraction, parameter measurement, image analysis, pattern recognition of individual pig. The accurate and effective image segmentation algorithm is for pig behavioral intelligent analysis, understanding and environmental intelligent decision. The traditional Graph Cut algorithm uses the energy function to compute all the pixels in the image, it takes a lot of time. It has the low segmentation efficiency and it cannot accurately segment specific pig. Aiming at the limitation of traditional Graph Cut algorithm and the characteristic of image itself, on the basis of the interactive watershed algorithm, in this paper, we proposed an interactive image segmentation method based on improved Graph Cut algorithm aiming at better use of interaction information provided by the user and effectively control of the number of super pixels. To some extent, the algorithm could also avoid the problems of over segmentation or under segmentation. The image was divided into regions based on the interactive watershed algorithm, and the foreground and background were specified. Each region block was regarded as the super pixel, the gray scale of super pixel was used as the vertex. Establish edges between adjacent vertices and the new network map was constructed instead of the traditional weighted graph. In order to reconstruct the energy function to complete the effective segmentation of the foreground and background, the maximum flow, and minimization cut algorithm were used to cut the image. Results of experiments showed that the method can effectively segment pig target, it had high segmentation quality and high efficiency. In this paper, the peak signal to noise ratio (PSNR) and structural similarity (SSIM) were used as objective evaluation standard. The range of the PSNR was [30, 40] and the range of the SSIM was [0.9, 1] in this method, which was the most accurate segmentation algorithm in the shown segmentation algorithm. It indicated that two evaluation criterions were consistent with the subjective evaluation, and the stability and the reliability of this method were improved. The average time consumed in this algorithm was only 33.7% of in the traditional Graph Cut algorithm. It greatly improved the efficiency of segmentation, and it had highly robust for the segmentation of specific target pigs. The results of this paper can provide technical support for subsequent target identification, detection, tracking and monitoring, evaluation of external environments, such as alarms. [ABSTRACT FROM AUTHOR]

In order to extract the crop rows of wheat quickly and accurately, a new method of wheat crop row detection was proposed based on particle swarm optimization (PSO) - clustering. The first step is image segmentation. The purpose of image segmentation was to separate the green plants from background, and it required the following 2 steps: Firstly, gray-level transformation, which could be done in RGB color space. In this paper, the color excess green index 2G-R-B was used; Secondly, image binarization was conducted. Among the global thresholding techniques for image binarization, Otsu method is one of the best threshold ways. So, this paper used Otsu algorithm to binarize the above obtained gray-level image. In order to reduce the burden of the next work, it was essential to extract a number of feature points indicating the crop rows. The specific algorithm can be divided into 2 steps: Firstly, get the left and right boundary points of each crop row. Secondly, extract the midpoint between left and right boundary points. After the original crop image was processed by the above steps, we got the feature points of the crop rows. According to the characteristics that the distances from the feature points around the crop row centreline to this straight line were all smaller than a certain distance threshold, we used the clustering method based on PSO to determine the real center points indicating crop rows. In the crop rows detection algorithm based on the PSO-clustering, the line in the data space composed of the feature points was considered as a particle. Finally, the centrelines were detected by fitting a straight line with the least square method. In order to prove the superiority of the algorithm, we compared the algorithm with standard Hough transform and the algorithm proposed in another literature. We tested the performance from the aspects of the detection accuracy and processing time for different images. Here, a total of 350 images have been tested. The number of the wheat images in overwintering stage was 197 and the number of the wheat images in green stage was 153. For the algorithm proposed in this paper, the number of the wheat images in overwintering stage successfully detected was 190 and that in green stage successfully detected was 143. Comparatively speaking, for the algorithm with standard Hough transform, the numbers of the wheat images in overwintering and green stage that were successfully detected were 180 and 100, respectively. For the algorithm proposed in another literature, the numbers were 168 and 93, respectively. Three representative pictures were selected in the experiment, which included the different environment i.e. the lack of crops, soil block, and high density weed. For the 3 images, the identification errors of the proposed algorithm were 0.631 0°, 0.773 5° and 1.065 7°, respectively. The identification errors of the standard Hough were 0.968 3°, 2.715 8° and 4.463 3°, respectively. The identification errors of the algorithm proposed in another literature were 2.260 5°, 1.531 9° and 5.829 1°, respectively. Therefore, compared with the other 2 algorithms, the proposed algorithm has the advantages of high real time and high accuracy, which can meet the practical requirements of field operation of agricultural robots. [ABSTRACT FROM AUTHOR]

In order to detect cutting surface roughness of grape grafting hard branch, this paper built a detection system of cutting surface roughness and designed an image detection algorithm of feature extraction based on measurement principle of light-section method. The detection system was made up of 9J light-section microscope, industrial camera and computer. Sampling length was an important role in the calculation of roughness. In order to make up for the shortage of single sampling length (1 420 μm), this paper applied image mosaic technology to images acquired by multiple sampling, and thus got a longer sampling length. Meanwhile, a method was proposed to product matching template automatically, which could avoid the manual operation and improve the efficiency of image mosaic when matching the reference image and target image based on gray level information. Then the image mosaic algorithm was tested and the results showed that: The average running time was increased by 1.104 s and the average sampling length was increased by 1131.77 μm when the number of mosaic image increased by a roughness feature image. Moreover, fuzzy set theory was applied to the process of gray-scale transformation and can effectively guarantee the integrity of the single edge of the image after segmentation. In this paper, Otsu algorithm was used to segment image. In order to filter the defeat profile due to catheter lumen and tracheid cavity contained by cutting surface itself, local pixels of the roughness feature binary image were operated by way of human-computer interaction. The pixels of the defect position in the foreground image were set to zero and not involved in the subsequent roughness calculation. A method was also proposed to extract the single side edge by scanning pixels of per column one by one, the first pixel whose value was not zero at each column belonged to edge pixels set. According to the measurement principle of light-section method, the corresponding position relation could be established between the single edge extracted from roughness feature image and roughness profile in one place of grape cutting surface. So the value of roughness height parameters Ra and Rz could be obtained by calculating the single edge. In order to verify the feasibility that the design of image detection algorithm could measure cutting surface roughness of grape hard branch, the comparative experiment was conducted between the detection system built and KEYENCE VK-200 laser microscope. The experimental results showed that the relative error of Ra was 6.73%, which was measured by the roughness image detection algorithm. Therefore, within the allowable range of measurement error, the image detection algorithm based on light-section method and designed to measure cutting surface roughness of grape hard branch has good feasibility, which can help provide technical support for the further research of the impact of cutting parameters on cutting surface roughness and grafting survival rate of grape hard branch. [ABSTRACT FROM AUTHOR]

The research on crop phenotype is one of the important measures in crop breeding. Crop breeding is a technology on selecting the good seeds by the crop phenotype. The farm remote sensing image analysis is a simple and effective method for rapid analysis of crop phenotype. However, the farm remote sensing image acquired by UAV (unmanned aerial vehicle) will be affected by the noise. In order to process and analyze the remote sensing image accurately, the remote sensing image should be first denoised. Typical image denoising algorithm of frequency domain is wavelet threshold algorithm. Wavelet transform can identify the singular signal accurately; however, the traditional wavelet can't effectively deal with multidimensional signal. For example, two-dimensional wavelet obtained from one-dimensional wavelet tensor only has horizontal and vertical directions, and the wavelet filter is isotropic, which is not sensitive to the image with more directions of the texture; while using the traditional wavelet threshold denoising algorithm to handle noise image, it is easy to make texture region blurry compared with the common wavelets. It is well known that the shear transformation has been introduced into the definition of the shearlet, which makes shearlet can represent more directions. In order to capture the directions of local geometric features of image, shearlet must be compactly suppressed. Anisotropic expansion scale transform can be applied in multi-resolution analysis. From coarse scale to fine scale approximation signal, namely in the approximation image in smooth regions using coarse scale, and in the approximation image with directions of texture using fine scale, the optimal approximation of singular curves can be achieved through the application of multi-resolution analysis method. Shearlet filter is anisotropic, and it is very sensitive to the direction of the texture; however, with the increase of noise standard deviation, after denoising, using shearlet algorithm generates artifacts easily. In brief, the denoising algorithm based on the total variation model has the advantage of edge preserving and noise reduction. In the application of shearlet for image denoising, we find that the shearlet with noise in the texture region can detect the texture direction, and can effectively remove the noise; but when the noise standard deviation is large, the application of shearlet algorithm in the image can cause the denoised image to tend to produce some artifacts, which will be identified as noise texture perhaps due to the shear wave in the smooth region. In order to solve the problem of discrete shearlet algorithm, this paper proposes a method based on the combination of shearlet and total variation model to eliminate the artifacts. First, select the generated Symmlet quadrature mirror filter, and ascertain anisotropic scale parameters and shear parameters. Perform discrete shearlet transformation to farmland remote sensing image with multi-resolution analysis. Wavelet coefficients can be obtained using the conventional method by anisotropic wavelet transformation. Wavelet coefficients are projected into wavelet of image. Hard threshold algorithm is used to handle noise coefficients. Wavelet coefficients are adopted to reconstruct farm remote sensing image. After using total variation model to smooth artifacts, the mirror extension is used to preprocess the image. In part of the experiment, the algorithm proposed in this paper is compared with the shearlet algorithm and total variation algorithm, and this algorithm gets a PSNR (peak signal to noise ratio) 1dB more than that from total variation model, and an iteration number of total variation less than that from direct total variation. In addition, not only can this algorithm effectively denoise farm remote sensing image noise and visual effects, but also can remove artifact better than the discrete shearlet transform algorithm. [ABSTRACT FROM AUTHOR]

The amount of wheat ears per unit area is an important parameter for assessing wheat yield and wheat planting density. Generally, phenotype parameters should be obtained by manual count technique which is time-consuming and needs great effort. Aiming at the traditional image segmentation method based on color feature, texture feature and Haar feature, in this paper, we proposed an improved K-means algorithm for estimating the number of wheat ears. This method established a direct mapping relationship from the low-level features of the image to the number of wheat ears in the image through the color feature so that the target did not need to be segmented or detected. This type of method was more suitable for complex lighting and dense wheat ears, and had higher computational efficiency than counting methods based on image segmentation. This method made full use of the color feature of wheat ear image, took the area feature of local region extracted from local region as the basis of wheat ear judgment, and used the number of sub-regions in clustering region as the estimation value of wheat ear number, thus avoiding the task of target detection and location, and greatly improving the accuracy of Wheat ear counting. According to classification results of K-means method, the wheat ear image was divided into three regions. The green area pixel value representing the wheat eat region was set to 255, the pixel value of other areas was set to 0, so the binary image of wheat ears was obtained. Most of the binarized areas in the image were wheat ears, and a small part of the binarized area that was too small or too large was caused by the leaves. Following the steps were used to count wheat ears: 1) extracting the connected regions of the binary image and labeling them; 2) calculating the area of each connected region, the area was represented by the number of pixels in the region; 3) using the following method to filter out the connected region where the area was too small or too large; 4) finally, the number of wheat ears was obtained by counting the retained binary regions. This method established a direct classification relation from the image color feature of the lower layer to the image containing wheat ears so that the image did not need to be segmented or detected. Based on color feature clustering, this method could estimate the number of wheat ears in local area of space, and it was more scalable without training. The average prediction precision of the total wheatear number in a wheatear image for 12 wheatear images was 94.69%. And the Statistical error of the total wheatear number was between 2.17% and 7.41%, the average statistical error was 5.31%. The statistical experiment results showed that the accuracy of this algorithm to wheat ear counting was better than the traditional method based on image color feature and texture feature. [ABSTRACT FROM AUTHOR]

The body condition scoring (BCS) is an important tool of assessment method on body condition for sow raising and management. It has been divided into 5 grades that from emaciated to overly fat grade and each grade had a score. The traditional method has negative effect on animals and farmers, and the process is complex with excessive contact. The body size and shape of sows are correlated with their reproductive performance but are difficult to measure manually. There is subjective uncertainty in the process of manual measurement. The Kinect’s 3D reconstruction technology was used to estimate and analyze the buttock shape of sows. A total of 108 images of Large White sows were manually acquired by Kinect camera during the feeding process at Feng Ning Animal Experimental Building in Chengde, Hebei Province of China, from July 25, 2017 to August 15, 2017. The hip height and hip width were measured by using tape and the back fat thickness was measured by using back fat measuring instrument. The hip height, hip width and area of buttock were obtained by analyzing the key points of 48 images. In order to obtain the measurement points of the livestock, several processing steps were taken, and the steps were as follow: 1) The sow stall was removed manually by Geomagic, and the target pig was acquired. In order to improve post-processing speed, the vertex culling algorithm was used to simplify the Three-dimensional model. 2) Since the models acquired were from different angles, the principal component analysis (PCA) was used to acquire new coordinate system. By using the geometrical relationship among the coordinate axes, standard measuring coordinate system was defined in this paper. 3) According to the geometric feature of the measurement points, the hip height, hip width and area of buttock were obtained. The results showed that the root mean square error of estimated body size was less than 2.1 cm, which meet the requirements of precision. The slice method was used to draw a curve at the highest point of the area of buttock based on the point cloud data. The least square fitting method was used to get the curve of hip contour. The hip radius of curvature was calculated by derivation formula. The results showed that the height-width ratio, area and radius of curvature of the sow’s hip had the correlation with the back fat thickness. The correlation coefficients were 0.567, 0.502 and 0.951 respectively. With radius of curvature as the main parameter, the sow body condition estimation model was built based on the experience of hip morphology. 60 images were selected for validation. By comparing the measured and the estimated values of back fat thickness, the maximum absolute error of back fat detection is 1.3 mm and the average relative error is 3.7%. The accuracy of body condition assessment was 91.7% compared with the traditional methods. All the results mentioned above indicate that this study provides a non-contact body condition assessment method based on 3D reconstruction technology and has certain application potential in the real livestock productive. [ABSTRACT FROM AUTHOR]

China is the largest country for cultivating kiwifruit, and Shaanxi Province provides the largest production, which accounts for approximately 70% of the production in China and 33% of the global production. Harvesting kiwifruit in this region relies mainly on manual picking which is labor-intensive. Therefore, the introduction of robotic harvesting is highly desirable and suitable. The fast and effective recognition of kiwifruit in the field under natural scenes is one of the key technologies for robotic harvesting. Recently, the study on kiwifruit recognition has been limited to a single cluster and multi clusters in the field have seldom been considered. In this paper, according to growth characteristics of kiwifruit grown on sturdy support structures, an RGB (red, green, blue) camera was placed around 100 cm underneath the canopy so that kiwifruit clusters could be included in the images. We proposed a kiwifruit image recognition system based on the convolutional neural network (CNN), which has a good robustness avoiding the subjectivity and limitation of the features selection by artificial means. The CNN could be trained end to end, from raw pixels to ultimate categories, and we optimized the critical structure parameters and the training strategy. Ultimately, the network was made up of 1 input layer, 3 convolutional layers, 2 sub-sampling layers, 1 full convolutional layer, and 1 output layer. The CNN architecture was optimized by using batch normalization (BN) method, which normalized the data distribution of the middle layer and the output data, accelerating the training convergence and reducing the training time. Therefore, the BN layers were added after the 1, 3 and 5th convolutional layer (Conv1, Conv3, and Conv5 layer) of the original LeNet network. The size of all convolutional kernels was 5×5, and that of all the sub-sampling layers was 2×2. The feature map numbers of Conv1, Conv3, and Conv5 were 6, 16 and 120, respectively. After manual selection and normalizing, the RGB image of kiwifruit was transferred into a matrix with the size of 32×32 as the input of the network, stochastic gradient descent was used to train our models with mini-batch size of 100 examples, and momentum was set as 0.9. In addition, the CNN took advantages of the part connections, the weight sharing and Max pooling techniques to lower complexity and improve the training performance of the model simultaneously. The network used rectified linear units (ReLU) as activation function, which could greatly accelerate network convergence. The proposed model for training kiwifruit was represented as 32×32-6C-2S-16C-2S-120C-2. Finally, 100 images of kiwifruit in the field (including 5918 fruits) were used to test the model, and the results showed that the recognition ratios of occluded fruit, overlapped fruit, adjacent fruit and separated fruit were 78.97%, 83.11%, 91.01% and 94.78%, respectively. The overall recognition rate of the model reached 89.29%, and it only took 0.27 s in average to recognize a fruit. There was no overlap between the testing samples and the training samples, which indicated that the network had a high generalization performance, and the testing images were captured from 9 a.m. to 5 p.m., which indicated the network had a good robustness to lightness variations. However, some fruits were wrongly detected and undetected, which included the fruits occluded by branches or leaves, overlapped to each other and the ones under extremely strong sunlight. Particularly, 2 or more fruits overlapped were recognized as one fruit, which was the main reason to the success rate not very high. This phenomenon demands a further research. By comparing with the conventional methods, it suggested that the method proposed obtained a higher recognition rate and better speed, and especially it could simultaneously identify multi-cluster kiwifruit in the field, which provided significant support for multi-arm operation of harvesting robotic. It proves that the CNN has a great potential for recognition of fruits in the field. [ABSTRACT FROM AUTHOR]

During the detecting and locating process of green tomatoes based on obtained visible images, problems such as shadow caused by uneven illumination source, occlusion of stems and leaves, and occlusion between fruits, need to be solved. In this study, a machine vision algorithm was put forward, which aimed to determine fruit location and size of green tomatoes. Normalized cross-correlation function (NCC) is a feature detection method using template matching. The algorithm firstly detected the potential location of green tomatoes through fast normalized cross-correlation function (FNCC). Then a gray histogram based classifier was used to classify if the location corresponded to green fruit. The histogram based classifier was built based on fruit areas and non-fruit areas of a certain size (30×30 pixels) extracted from the obtained image. Seven texture features, including mean, standard deviation, smoothness, third moment, uniformity, entropy, and gray level range were calculated for those fruit areas and non-fruit areas. Three classifiers including k-nearest neighbor (KNN), SVM (support vector machine), and Naive Bayes, were used to classify fruit and non-fruit areas using those 7 texture features as input vectors. SVM was chosen based on its performance. The non-fruit location was filtered out, and the number of fruit locations was estimated. Meanwhile, the image was segmented based on color analysis. Red and Blue component from RGB (red, green, blue) image, and Hue component from HSV (hue, saturation, value) image transformed from RGB images, were used as the basis for color analysis. Using the fruit potential location number estimated using FNCC as an input parameter of circular Hough transform (CHT), CHT was then applied to the edge image of the segmented result. The center coordinates and radius value of each circle were calculated. Finally, the detection results were merged based on the analysis of the distances of 2 centers of fruit circles detected using CHT. If the distance between 2 circles is smaller than the minimum fruit radius, the circle with a larger radius will be kept, while the other circle will be flagged as repeatedly detected one. Thus, the recognition and positioning of the green tomato were realized. When green fruits and red fruits appear on the same image, a red fruit detection algorithm based on the local maximum value method and random circle round transform detection, which was proposed by the author in another paper, would be carried out on the obtained image. Then the red fruit detection result was combined with the green fruit detection result. The proposed method combined the texture, color, and shape information of the image, and eliminated the disturbance of the color similarity between the green tomatoes and green leaves and stems. A total of 70 images were used in this study, in which 35 images were used as training images, and the other 35 images were used as validation images. The correct detection ratio for 72 fruits in the training dataset was 89.2%, and the correct detection ratio for 105 fruits in the validation dataset was 86.7%. The proposed method has provided a reference for the development of tomato harvesting robots for both red and green mature tomato fruits. [ABSTRACT FROM AUTHOR]

Plant phenotype is a key link in understanding gene function and environmental effects, and with the demand of further research of plant function genomics, the traditional phenotypic observation has become the main bottleneck. High-throughput plant phenome analysis technology such as image processing is an effective way to solve this problem. In this paper, a maize plant drought stress phenotype testing method was proposed based on the time-series images. Firstly, an acquisition system of time-series images was developed. System hardware included the mvc3000 high speed industrial camera (the maximum resolution 1600×1200 pixels, 24 FPS (frames per second)), the Pentax len (8.5 mm, f/1.5), the storage card (32G), the holder and the background. Software development environment included Win7 operating system, Microsoft Visual Studio 2010 Professional, and OpenCV2.4.3. Maize plant time-series images were acquired from 08:00 to 17:00 every 10 mins, and various natural light conditions strongly affected the profile of the crop images taken outdoors. This often made it rather difficult for vegetation parts segmented from the background parts in an image. We adopted the K-means and SVM (support vector machine) algorithm to create a classifier used in rough segmentation, and the classifier was composed of 24 dimension feature vectors including color features, texture features, and gradient features. The result refinement was implemented with a non-local mean (NLM) filter, and actually the anisotropic structural-aware filter was adopted to implement the NLM function, in which the pixels were reinforced with each other in the same structure and even if they were not the first-order neighboring pixels. Secondly, the plant organs such as stem and leaf were segmented in binary image with a scan line method, which used the plant morphology characteristics such as alternate phyllotaxis, curved blade, and leaf margin in folds as prior knowledge. The method denoted the pixels according to the number of neighborhoods. Specifically, each row of pixels in the scan line was marked as a candidate region, and the candidate regions were labeled as the original organ or new organ according to the number of neighborhoods. When all of the candidate regions were labeled, a cluster of regions which presented rectangular shape were defined as plant stem. A labeled cluster of leaf would be merged to the other cluster if they shared neighborhood regions up to 70%. Finally, the ratio SOP1/SOP2 was used to characterize the drought stress response. SOP is the intersection point of stem and leaf, SOP is a point in stem line, LEN is the length of leaf tip to stem line, SOP1 is a point in leaf curve with length 1/4LEN, SOP2 is a point in leaf curve with length 2/3 LEN. SOP1/SOP2 represented the degree of leaf bending. It was comparable even between the different leaf orders, and had strong robustness to environmental factor disturbance. The field tests showed that the SOP1/SOP2 of drought-resistant genetically-modified maize presented an ascending trend before midday and a descending trend after then, which could be caused by transpiration leading to leaf water loss. The change trend of environmental factors such as air relative humidity against SOP1/SOP2 showed that under drought stress condition, genetically-modified maize was more sensitive to moisture content of the atmosphere, and the test on conventional maize plant showed that the SOP1/SOP2 lacked the regularity with the time. [ABSTRACT FROM AUTHOR]

The Kinect sensor scanning images for agricultural plants are vulnerable to field light conditions and background noise, etc. In addition, the amount of data of the color image and depth image affect the efficiency and accuracy of the plant area, which leads to the difficulty of meeting the requirement for the Kinect sensor in agricultural plant detection. For the above problems, considering the influence of the light conditions and complex background information in agricultural environment on the quality of the plant detection and the depth data acquisition, in this paper, we proposed a plant depth detection image restoration method based on K-means and K-nearest neighbor. We also developed a novel method of image restoration to reduce the impact of background information to improve the accuracy of the color image segmentation, and to enhance the accuracy of depth data. Firstly, a RGB threshold segmentation algorithm was applied to original RGB-formatted plant color images to extract plant target areas from backgrounds. Three components R, G, and B were respectively separated from RGB color space, and the difference between G and R or B was primary extract of the plant area information. Meanwhile, for the color characteristic of the environment, a K-means clustering segmentation algorithm was performed on the extracted plant target areas to remove background noise and enhance target contours. Secondly, to fix the errors of the depth data and meet the requirements of the agricultural plant detection operations, the color image and depth image were registered to restore the suspicious pixels depth data based on K-nearest neighbor algorithm. Then, a K-Nearest Neighbor algorithm was presented to recovery the black hole pixels for depth images. Finally, we acquired the depth data of target plant from the detected images. Compared with conventional RGB threshold segmentation method and K-means algorithm method, the proposed method can be used to solve the problem of the color image noise. The experiment results showed that, the segmentation error can be reduced by 12.12% with RGB threshold segmentation method, and 41.48% with K-means algorithm method. The average segmentation error can be up to 12.33% by using RGB threshold segmentation first and then the K-means algorithm. Furthermore, the proposed method can be used to restore the depth data, and can significantly reduce the effect of the backgrounds. Thus it had a good improvement to the edge sharpness of the depth data, and the accuracy of the empty point depth data of single frame. The result of this study can be a reference for agricultural plant detection and 3D reconstruction, precision of target spraying. [ABSTRACT FROM AUTHOR]

Image processing and analysis play the key roles in the smart agriculture system. As the agricultural images are often taken in the open air, the heavy atmospheric haze and dust in the air are often included in the images. It’s very difficult to differentiate the noises and the detail texture in the images using existing common methods. As the detail texture is often employed to identify the important feature of different plants, the texture preserving image denoising algorithm plays an important role in the agricultural field. The existed methods can’t identify the detail texture and the noise, and so it is difficult to fulfill the texture preserving image denoising. The shearlet transform proposed in recent years possesses the multi-scale directivity as the shear matrix is introduced into it. So the shearlet transform can be used to describe the detail texture precisely. However, the shortcoming of the shearlet transform is that it often mistakes the noise as the texture in many cases. On the contrary, the variational method views all the objects in the images as the smooth domains, this results in detail texture in the images being often destroyed. To solve the problem, we proposed a coupling technology of the multi-scale variational method based on the interpolation wavelet frame and the shearlet transform, in which the variational method was employed to identify the contour and the shearlet transform to describe the texture precisely. According to this method, the image was firstly decomposed and reconstructed by means of the shearlet transform, which can remove most of the noise in the image. Second, the variational method based on the multi-scale interpolative wavelet frame was employed to smooth the denoised image. This can divide the image into some domains, and they possessed different texture feature which could be identified by means of the correlation value derived from the gray-level co-occurrence matrix of the grayscale image. Compared to the variational method under the total variation (TV) frame, the multi-scale interpolative wavelet frame can identify more detail domains which was helpful to improve the quality of the denoised images. To the image noised by the Gauss noise with the standard derivation 10, the peak signal to noise ratio (PSNR) of the denoised image obtained by the variational method based on the multi-scale interpolative wavelet frame was 1.3131 larger than PSRN obtained by one based on the TV frame, and the structural similarity image measurement (SSIM) increased by 4.5% accordingly. Next, the variational method and the shearlet transform method can be used to remove the noises existed in the cartoon region and the texture region, respectively. This can overcome the shortcomings of only one method and remain all the merits in all the denoising algorithms. For instance, median filtering can be applied to remove the noise in the smooth domain, and the shearlet transform can be employed to remove the noise in the domain with abundant textures. Last, the locust slice images were taken as the examples to illustrate the proposed method. The numerical experiments showed that coupling technology proposed in this paper can improve the image denoising precision effectively compared to each single method. The value of the PSNR was improved by 6.37% compared to the multi-scale variational method, and by 5.90% compared to the shearlet transform method. Compared with another attractive algorithm K-SVD which is an excellent algorithm for designing of overcomplete dictionaries for sparse representation, the coupling technology also had apparent superiority. To the locust slice images noised by the Gauss noise with the standard derivation 40, PSRN obtained by the coupling technology increased by 0.43 compared to the K-SVD algorithm, and SSIM was improved by 2.5% accordingly. In addition, the coupling technology was independent of the images, and the K-SVD algorithm yielded sparse representation for the training signals which limits its application to some extent. [ABSTRACT FROM AUTHOR]

Using real-time video to capture farmland digitization regulation is an important step in the protection of essential farmland, but problems exist such as a large area of bright sky background and extreme weather i.e. image degradation caused by rain or fog. Generally, we run the defogging process which is used for image processing to clarify the hazy image in the service. However, the cost is high and the process is in unreal time so that it is not suitable for the storage of video data and real-time alarm. With the innovation of computer hardware, it is possible now to defog in real-time under the haze weather. The Langley research center of the national aeronautics and space administration transplanted the algorithm which is based on Retinex algorithm to DSP(Digital signal process) enhancement system that meets the real-time requirements to deal with 256×256 gray-scale image. ClaireVue’s team from Tsinghua University developed a real-time system on the iPhone 4 to defog 192×144 video image. Cai Zixing’s team from Central South University put forwarded an algorithm based on the mist theory to achieve theoretical efficiency of real-time processing. In this paper, we aim at defogging the basic farmland video surveillance images in real-time. We achieved the MDCP(modified dark channel prior) algorithm which was improved on the basis of dark channel prior defogging algorithm with the combination of the dividing and merging of human visual perception to hazing. We built up a system which can clarify the basic farmland video surveillance image by using the subsample of transmittance, adjacent pixels completion, application processing block and the front-end hardware layered method to defog. In order to objectively demonstrate the effectiveness of MDCP algorithm, we used no reference evaluation model to evaluate MDCP algorithm, dark channel prior algorithm and multi-scale Rentinex algorithm and we gained the objective assessment of clarifying the hazy image. The data showed that MDCP algorithm was more prominent in intensity, tone reproduction and information of structural aspects than the other two algorithms in clarification. MDCP algorithm had the highest comprehensive evaluation indicators following by DCP(dark channel prior), and Retinex was the lowest among the three. The defogging system which was used for basic farmland video surveillance included video input device(camera), real-time image processing device(DSP hardware system) and monitor. We adopted SONY SSC-G103 CCD(Charge-couple Device) camera as video input device. DSP hardware system which is made up by the TMS320DM642 platform helped to complete the acquisition of video codec, to format conversion, and to clarify processing. Our test which was based on video image processing and algorithm complexity showed that the system improved the strength of hazy image, tone reproduction and structural information indicator and it kept the good real-time and fluency. In summary, the basic farmland video monitoring front-end of defogging system had some advantages of low cost, low power consumption and processing in real-time while comparing to traditional system which enhanced image in services. It achieved the goal that defogging the video surveillance image in front-end and real-time. [ABSTRACT FROM AUTHOR]

Detection of pork backfat thickness in most of the slaughtering houses depends on manual labors using measuring tools. The objective of this research was to investigate the method for detecting backfat thickness based on computer vision and image processing technologies. And the paper proposed an algorithm of image acquisition and automatically measuring backfat thickness which could solve the problems that manual measurement process had low efficiency, human factor influenced the test result and connective tissue was readily measured as backfat region. The images of pig carcass between the 6th and the 7th rib were collected by the machine vision image acquisition system on the slaughter line. The system consisted of an image acquisition module containing CCD (charge-coupled device) to capture the images and then save them in computer, a single-chip microcomputer, a detection switch, the calibration rule and the light source in system that could be regulated by the controller to change intensity, and the image processing algorithm was equipped into the self-developed system embedded in the computer. The distance between the camera lens and the carcass samples was fixed. A black background plate was placed behind the pig carcass in order to adapt to the complexity of the environment. When a half of carcass reached the camera view, the operator pressed the detection switch to acquire images which were automatically stored in the computer for further image processing. First, the image noise was removed by using the bilateral filtering method. And the binary image of the pig carcass to be detected was gained according to the Otsu method which calculated segment threshold automatically based on the image grey value. After filling the tiny holes in the binary images by using morphological transformation, the images still contained multiple connected regions. Then the image contours were extracted from the preprocessed images. Through the experiment, it was found that the backfat region was the largest region in the image contour region. Based on the differences of different contour sizes, the backfat region and edge contour were obtained. Secondly, the edge contours were fitted by the fitting line to yield the standard deviations, which were then used to determine whether the connective tissue existed in the backfat region. If so, the pixels of the backfat region image accumulated along X direction were plotted. The connective tissue was removed using the new detection line determined by the valley point coordinates of pixel curve. In this step, the image was cropped to separate the backfat region from the original image. Finally the backfat thickness could be measured accurately by mapping the line to the backfat region. Experiment showed that the detection accuracy of measuring the backfat thickness was 93.5% when the measurement error was less than 1 mm. The accuracy of the algorithm and the speed were verified with the theoretical analysis and practical test. And through test, the average recognition time of each sample was 0.3 s. The results showed that the algorithm could meet the requirement of the backfat thickness testing and measuring in precision for the practical application. This method is able to be used in online detection of the slaughtering line which is of great significance for the development of the automatic measuring system. [ABSTRACT FROM AUTHOR]