Search

Abstract With the popularity of social network, social media data embedding has attracted extensive research interest and boomed many applications, such as image classification and cross-modal retrieval. In this paper, we examine the scenario of social images containing multimodal content (e.g., visual content and textual tags) and connecting with each other (e.g., two images submitted to the same group). In such a case, both the multimodal content and link information provide useful clues for representation learning. Therefore, simply learning the embedding from network structure or data content results in sub-optimal social image representation. In this paper, we propose a Deep Multimodal Attention Networks (DMAN) to combine multimodal content and link information for social image embedding. Specifically, to effectively incorporate the multimodal content, a visual-textual attention model is proposed to encode the fine-granularity correlation between multimodal content, i.e., the alignment between image regions and textual words. To incorporate the network structure for embedding learning, a novel Siamese-Triplet neural network is proposed to model the first-order proximity and the second-order proximity among images. Then the two modules are integrated into a joint deep model for social image embedding. Once the representation has been learned, a wide variety of data mining problems can be solved by using the task-specific algorithms designed for handling vector representations. Extensive experiments are conducted to demonstrate the effectiveness of our approach on multi-label classification and cross-modal search. [ABSTRACT FROM AUTHOR]

This paper combines the discriminative feature extraction and effective classifier construction into a single framework to learn a structured discriminative dictionary for image classification. Due to the fact that the discriminative signal lie in a low dimensional subspace and can be well represented only via a few atoms of the learned dictionary, this paper addresses the feature extraction via learning a dictionary, whose sub dictionaries preserve correspondence to the class labels, and an optimal linear classifier jointly based on the structure of energy contribution. Based on the discriminative energy contributions, we are searching the discriminative feature for classification rather than reconstructing the data accurately. In addition, with the assumption that the classifier has a specific property which is similar with the dictionary, we learn a classifier to make the dictionary optimal and have a low cost on classifying. Experiment results on the several databases to specific classification tasks are conducted to verify the efficacy of the proposed method compared with the state-of-the-art dictionary learning for classification methods. [ABSTRACT FROM AUTHOR]

Abstract Most of existing image categorizations assume that the given datasets have a good resolution and quality. However, the assumption is often violated in real applications. In this paper, we study the low-resolution (LR) image categorization. By utilizing labeled high-resolution (HR) images as auxiliary information, we formulate the problem as a heterogeneous domain adaptation problem and propose a Discriminative Joint Distribution Adaptation (DJDA) model to solve it. The DJDA model projects both LR and HR images into an intermediate subspace with a well-designed objective function, where the distance between classes is expected to be enlarged and the distribution divergence to be reduced. As a result, the discriminative knowledge for HR images can be transferred effectively to LR images. Experimental results demonstrate the proposed DJDA method produces significantly superior categorization accuracies against state-of-the-art competitors. [ABSTRACT FROM AUTHOR]

Abstract Two Phase Test Sample Sparse Representation (TPTSSR) classifier was recently proposed as an efficient alternative to the Sparse Representation Classifier (SRC). It aims at classifying data using sparse coding in two phases with ℓ 2 regularization. Although high performances can be obtained by the TPTSSR classifier, since it is a supervised classifier, it is not able to benefit from unlabeled samples which are very often available. In this paper, we introduce a semi-supervised version of the TPTSSR classifier called Semi-supervised Two Phase Test Sample Sparse Representation (STPTSSR). STPTSSR combines the merits of sparse coding, active learning and the two phase collaborative representation classifiers. The proposed framework is able to make any sparse representation based classifier semi-supervised. Extensive experiments carried out on six benchmark image datasets show that the proposed STPTSSR can outperform the classical TPTSSR as well as many state-of-the-art semi-supervised methods. [ABSTRACT FROM AUTHOR]

Multi-level image thresholding is a common approach to image segmentation where an image is divided into several regions based on its histogram. Otsu's method is the most popular method for this purpose, and is based on seeking for threshold values that maximise the between-class variance. This requires an exhaustive search to find the optimal set of threshold values, making image thresholding a time-consuming process. This is especially the case with increasing numbers of thresholds since, due to the curse of dimensionality, the search space enlarges exponentially with the number of thresholds. Population-based metaheuristic algorithms are efficient and effective problem-independent methods to tackle hard optimisation problems. Over the years, a variety of such algorithms, often based on bio-inspired paradigms, have been proposed. In this paper, we formulate multi-level image thresholding as an optimisation problem and perform an extensive evaluation of 23 population-based metaheuristics, including both state-of-the-art and recently introduced algorithms, for this purpose. We benchmark the algorithms on a set of commonly used images and based on various measures, including objective function value, peak signal-to-noise ratio, feature similarity index, and structural similarity index. In addition, we carry out a stability analysis as well as a statistical analysis to judge if there are significant differences between algorithms. Our experimental results indicate that recently introduced algorithms do not necessarily achieve acceptable performance in multi-level image thresholding, while some established algorithms are demonstrated to work better. [ABSTRACT FROM AUTHOR]

Recently, a large number of geo-tagged landmark images have been uploaded through various social media services. Usually, these geo-tagged images are annotated by users with GPS and tags related to the landmarks where they are taken. Landmark tagging aims to automatically annotate an image with the tags to describe the landmark where the image is taken. It has been observed that the images and tags show strong correlation with the geographical locations. The widely used assumption by many existing tagging methods is that images are independently and identically distributed is not effective to capture the geographical correlation. In this paper, we study the novel problem of utilizing the geographical correlation among images and landmarks for better tagging landmark images. In particular, we propose an unsupervised feature learning approach to learn the geographically discriminative features across geographical locations, by integrating latent space learning and geographically structural analysis (LSGSA) into a joint model. A latent space learning model is proposed to effectively fuse the heterogeneous features of visual content and tags. Meanwhile, the geographical structure analysis and group sparsity are applied to learn the geographically discriminative features. Then, a geo-guided sparse reconstruction method is proposed to tag images by utilizing the discriminative information of features, in which the landmark-specific tags are boosted by a weighting method. Experiments on the real-world datasets demonstrate the superiority of our approach. [ABSTRACT FROM AUTHOR]

Most of the available pan-sharpening methods use vector or matrix based detail injection to enhance the resolution of MultiSpectral (MS) image, which may result in unavoidable spectral and spatial distortions. In this paper we explore the intrinsic tensor structure and local sparsity of MS images, to develop a novel Sparse Tensor Neighbor Embedding (STNE) based pan-sharpening method that reduces the distortions in the fused images. First, MS images are formulated as some spectral tensors, and each tensor and its nearest neighbor tensors are assumed to lie in a low-dimensional manifold. Then the tensor is sparsely coded under its neighbor tensors, and a joint sparse coding assumption is cast on bands to develop an N-way Block Pursuit algorithm for solving sparse tensor coefficients. Finally high resolution MS tensor can be obtained by weighting Panchromatic image with the sparse tensor coefficients. Tensors are higher order generalizations of vectors and matrices, and taking advantage of high-order structure of multi-dimensional data can help us understand them. The proposed method first combines a sparse tensor with neighbor embedding, to construct a new high-dimensional sparse tensor embedding for efficient pan-sharpening. Because tensor formulation can exploit the structural correlations in high-dimensional MS data, the proposed method can well preserve spectral correlation among different bands simultaneously and capture the underlying high-order statistical properties of MS image. Some experiments are performed on several real QuickBird and GeoEye datasets, and experimental results show that STNE is superior to its counterparts in reducing spectral and spatial distortions. [ABSTRACT FROM AUTHOR]

Multiple image variations occur in natural face images, such as the changes of pose, illumination, occlusion and expression. For non-specific variations based face recognition, learning effective features is an important research topic. Subspace learning is a widely used face recognition technique; however, numerous subspace analysis methods do not fully utilize the prior information of facial variations. Tensor-based multilinear subspace analysis methods can take advantage of the prior information, but they need to be further improved. With respect to a single facial variation, we observe that the image samples belonging to the same variation-state but different classes tend to cluster together, whereas those belonging to different variation-states but the same class tend to remain separate. This is adverse to classification. In this paper, motivated by the idea of charge law, “like charges repulsion and opposite charges attraction”, in which like and opposite charges are regarded as same and different variation-states, respectively, we propose a non-specific variations based discriminant analysis (NVDA) criterion. It searches for an optimal discriminant subspace in which samples belonging to same variation-state but different classes are separable, whereas those belonging to different variation-states but same class cluster together. We then propose a novel face recognition approach called non-specific variations based multi-subspace analysis (NVMSA), which serially utilizes NVDA criterion to learn multiple discriminant subspaces corresponding to different variations. In the proposed approach, we design a strategy to select the serial calculation order of variations and provide a rule to choose projection vectors with favorable discriminant capabilities. Furthermore, we formulate the locally statistical orthogonal constraints for the multiple subspaces learning to remove the local correlation of discriminant features obtained from multiple variations. Experiments on the AR, Weizmann, PIE and LFW databases demonstrate the effectiveness and efficiency of the proposed approach. [ABSTRACT FROM AUTHOR]

Data mining has great potential in extracting useful knowledge from large amount of temporal data for dynamic decision-making. Moreover, integrating visualization in data mining, known as visual data mining, allows combining the human ability of exploration with the analytical processing capacity of computers for effective problem solving. To design and develop visual data mining tools, an appropriate process must be followed. In this context, the goal of this paper is to enhance existing visualization processes by adapting it under the temporal dimension of data, the data mining tasks and the cognitive control aspects. The proposed process aims to model the visual data mining methods for supporting the dynamic decision-making. We illustrate the steps of our proposed process by considering the design of the visualization of the temporal association rules technique. This technique was developed to assist physicians to fight against nosocomial infections in the intensive care unit. Actually, an evaluation study in Situ was performed to assess the automatic prediction results as well as the visual representations. At the end, the test of the efficiency of our process using utility and usability evaluation shows satisfactory. [ABSTRACT FROM AUTHOR]

Recently, ConvNeXts constructing from standard ConvNet modules has produced competitive performance in various image applications. In this paper, an efficient model based on the classical UNet, which can achieve promising results with a low number of parameters, is proposed for medical image segmentation. Inspired by ConvNeXt, the designed model is called ConvUNeXt and towards reduction in the amount of parameters while retaining outstanding segmentation superiority. Specifically, we firstly improved the convolution block of UNet by using large convolution kernels and depth-wise separable convolution to considerably decrease the number of parameters; then residual connections in both encoder and decoder are added and pooling is abandoned via adopting convolution for down-sampling; during skip connection, a lightweight attention mechanism is designed to filter out noise in low-level semantic information and suppress irrelevant features, so that the network can pay more attention to the target area. Compared to the standard UNet, our model has 20% fewer parameters, meanwhile, experimental results on different datasets show that it exhibits superior segmentation performance when the amount of data is scarce or sufficient. Code will be available at https://github.com/1914669687/ConvUNeXt. [ABSTRACT FROM AUTHOR]

Spatial Pyramid Matching (SPM) and its variants have achieved a lot of success in image classification. The main difference among them is their encoding schemes. For example, ScSPM incorporates Sparse Code (SC) instead of Vector Quantization (VQ) into the framework of SPM. Although the methods achieve a higher recognition rate than the traditional SPM, they consume more time to encode the local descriptors extracted from the image. In this paper, we propose using Low Rank Representation (LRR) to encode the descriptors under the framework of SPM. Different from SC, LRR considers the group effect among data points instead of sparsity. Benefiting from this property, the proposed method (i.e., LrrSPM) can offer a better performance. To further improve the generalizability and robustness, we reformulate the rank-minimization problem as a truncated projection problem. Extensive experimental studies show that LrrSPM is more efficient than its counterparts (e.g., ScSPM) while achieving competitive recognition rates on nine image data sets. [ABSTRACT FROM AUTHOR]

Noise components are a major cause of poor performance in document analysis. To reduce undesired components, most recent research works have applied an image processing technique. However, the effectiveness of these techniques is suitable only for a Latin script document but not a non-Latin script document. The characteristics of the non-Latin script document, such as Thai, are considerably more complicated than the Latin script document and include many levels of character alignment, no word or sentence separator, and variability in a character’s size. When applying an image processing technique to a Thai document, we usually remove the characters that are relatively close to noise. Hence, in this paper, we propose a novel noise reduction method by applying a machine learning technique to classify and reduce noise in document images. The proposed method uses a semi-supervised cluster-and-label approach with an improved labeling method, namely, feature selected sub-cluster labeling. Feature selected sub-cluster labeling focuses on the clusters that are incorrectly labeled by conventional labeling methods. These clusters are re-clustered into small groups with a new feature set that is selected according to class labels. The experimental results show that this method can significantly improve the accuracy of labeling examples and the performance of classification. We compared the performance of noise reduction and character preservation between the proposed method and two related noise reduction approaches, i.e., a two-phased stroke-like pattern noise (SPN) removal and a commercial noise reduction software called ScanFix Xpress 6.0. The results show that semi-supervised noise reduction is significantly better than the compared methods of which an F-measure of character and noise is 86.01 and 97.82, respectively. [ABSTRACT FROM AUTHOR]

Speed management represents an important strategy in order to improve road safety, because a strong relationship is between speed and crash occurrence and severity. Speed limits enforcement is the main measure to control operating speeds but to obtain the compliance of drivers, the limits must be safe but also credible. This means that road users have to regard the speed limit as logical under given conditions and so a speed limit is credible if it fits the image evoked by the road. This paper describes the development of a Decision Support System (DSS) for the selection of safe and credible speed limits for speed zones. The proposed DSS is based on Dominance-based Rough Set Approach (DRSA), which presents interesting advantages in terms of transparency and manageability with respect to many other decision support competitive methodologies. In fact DRSA, after getting the preferred information necessary to set up the decision model, in terms of exemplary decisions, allows to build a multi-criteria model expressed in terms of “if …, then …” decision rules. The proposed multi-criteria decision approach aims to suggest to decision-makers a safe and credible speed limit for speed zones, taking into consideration many factors such as accident rate, roadway geometry, roadway development, traffic and others. [ABSTRACT FROM AUTHOR]

Flame image recognition is of great significance in the fire detection and prevention. In this paper, in order to improve the accuracy of fire recognition, a fast stochastic configuration network (FSCN) method based on an improved sparrow search algorithm (ISSA) is proposed. In the design of fast stochastic configuration network (FSCN), the gradual increase of hidden layer nodes in the original stochastic configuration network (SCN) is canceled, and the number of them is set directly. An improved sparrow search algorithm (ISSA) is used to generate the input weights and biases of hidden layer nodes. At the same time, the supervisory mechanism is retained to judge the weights and biases of all hidden layer nodes, and ISSA is used to regenerate corresponding weights and biases for the nodes that do not meet the constraints in the supervisory mechanism. In the ISSA, sine map, adaptive adjustment of hyper-parameters and mutation strategy are used to improve the optimization ability of the original sparrow search algorithm (SSA). Some parameters in FSCN are optimized by ISSA to make it have better classification performance. Finally, the image processing technology is used to extract features from the flame images and the interference images, and then the feature vectors are obtained to train the ISSA-FSCN. Several simulation experiments have been carried out to verify effectiveness of the proposed ISSA-FSCN method. In the performance verification of ISSA on CEC test suit, ISSA averagely outperforms other algorithms by 33.6% in the average results of 20 functions. In the performance verification of FSCN, the average results of a c c u r a c y , p r e c i s i o n , r e c a l l , F 1 and a u c are compared. In the experiment 1, ISSA-FSCN averagely outperforms other algorithms by 19.7%, 14.7%, 12.8%, 14.5% and 23.0%. In the experiment 2, ISSA-FSCN averagely outperforms other algorithms by 2.3%, 2.1%, 2.5%, 6.0%, 3.2%. In the experiment 3, ISSA-FSCN averagely outperforms other algorithms by 5.9%, 4.0%, 4.1%, 4.1%, 6.8%. • A fast stochastic configuration network (FSCN) was proposed. • An improved sparrow search algorithm was proposed with good optimization ability. • The proposed ISSA-FSCN was applied to the classification of flame images. [ABSTRACT FROM AUTHOR]

Image segmentation has been an important technique in the field of image processing. Fine-level manual annotations are very limited and difficult for a large collection of imbalanced images, where each image contains quite different small objects. However, we find that these imbalanced images have similar decision graphs obtained by a lightweight and simple clustering algorithm Density Peaks (DPeaks). Hence, in this paper, we propose a weakly supervised image segmentation algorithm. It trains a decision curve from decision graphs of a few sample imbalanced images by SVM and Support Vector Regression (SVR), which is effective for identifying the sparse region of an imbalanced image. Besides, RangeTree is applied to accelerate RDP for large images due to the high complexity of DPeaks Clustering. Experiments prove that the proposed algorithm works well on imbalanced image data sets, not only it can recognize main things, but also has the ability to identify some relatively small objects. [ABSTRACT FROM AUTHOR]

Low-rank representation (LRR) can recover clean data from noisy data while effectively characterizing the subspace structures between data, therefore, it becomes one of the state-of-the-art methods for subspace learning and is widely used in machine learning, image processing, and data mining. In this paper, we propose a novel three-term low-rank tensor decomposition approach called the enhanced tensor LRR (ETLRR). In ETLRR, the original data tensor is decomposed into three parts: low-rank structure tensor, sparse noise tensor, and Gaussian noise tensor. First of all, unlike the existing LRR-related methods, which only consider one kind of Laplacian or Gaussian distribution noise, we consider that two types of noise can effectively restore a clean tensor, thereby obtaining a more accurate low-rank tensor subspace structures. Secondly, the denoised tensor rather than the original data tensor is adopted to construct the dictionary. And then, ETLRR can be implemented directly on the tensor data composed of the samples while two-dimensional data such as image samples are not converted into vectors in advance. Finally, we propose an iterative update method for the optimization of ETLRR based on the alternating direction method of multipliers (ADMM). Compared with the state-of-the-art methods, experiments on synthetic data and image clustering, image and video denoising verify the good performance of ETLRR in both obtaining the low-rank tensor subspace structures and recovering the tensor data. [ABSTRACT FROM AUTHOR]

Thresholding is a direct and simple approach to extract different regions from an image. In its basic formulation, thresholding searches for a global value that maximizes the separation between output classes. The use of a single hard threshold value is precisely the source of important segmentation errors in many scenarios like noisy images or uneven illumination. If no connectivity or closed objects are considered, the method is prone to produce isolated pixels. In this paper a new multiregion thresholding methodology is presented to overcome the common drawbacks of thresholding methods when images are corrupted with artifacts and noise. It is based on relating each pixel in the image to different output centroids via a fuzzy membership function, avoiding any initial hard decision. The starting point of the technique is the definition of the output centroids using a clustering method compatible with most thresholding techniques in the literature. The method makes use of the spatial information through a local aggregation step where the membership degree of each pixel is modified by local information that takes into account the memberships of the surrounding pixels. This makes the method robust to noise and artifacts. The general formulation of the proposed methodology allows the design of spatial aggregations for multiple applications, including the possibility of including heuristic information via a fuzzy inference rule base. [ABSTRACT FROM AUTHOR]

Considering that saliency areas in images play important roles in human visual perception, in this paper we propose a Learning Compressive Sampling (LCS) scheme and its physical implementation for more efficient image acquisition via a new Multiscale and Steerable Support Value Transform (MS2VT). The key idea is to learn geometric saliency maps of images by MS2VT, which is deduced from a mapped least square-support vector machine. Because MS2VT can produce a multiscale, multidirectional, undecimated, dyadic and aliasing transform with shift-invariant and anisotropy properties, the obtained support values can reveal the geometric and saliency information of images. The learned saliency map is then used to formulate a variable density compressive function, to realize a simple, fast and efficient sampling, which aims to allocate more sensing resources to saliency attention areas but fewer to non-salient regions. Several experiments are taken on some natural images and remote sensing images to compare our proposed LCS scheme with traditional samplings when saliency information is not used. Moreover, the performance of MS2VT based saliency detection scheme is also compared with other related saliency detection approaches. The experimental results indicate that it can obtain high quality images, especially in preserving more detailed edges, contours and complex structures, even at low sampling ratios. [ABSTRACT FROM AUTHOR]

Computer vision is an interdisciplinary field that includes methods for acquiring, processing, analyzing, and understanding visual information. In computer vision, usually the approaches to solve problems are specific-application methods and, therefore, reusing captured knowledge in computer vision is hard. However, the aim of knowledge modeling (KM) is to capture and reuse knowledge to solve different problems. In this paper, we propose a knowledge-based component library for computer vision tasks such as video surveillance applications. The developed components are based on the region of interest (ROI) a well-known concept in the image processing and computer vision fields. We provide a set of reusable components that are specializations and/or compositions of ROIs. Finally, we propose several case studies that illustrate the feasibility of the proposal. Experimental results show that the proposed method deals effectively and efficiently with real-life computer vision problems. [ABSTRACT FROM AUTHOR]

Existing research works in scene image classification have focused on different aspects such as content features (e.g., visual information), context features (e.g., annotations, semantic information, etc.) and both. However, such works suffer from various issues such as higher feature size and lower classification performance. In this paper, we propose a new feature extraction approach for scene image representation using two kinds of rich information: content features and context features. Specifically, the new content features are generated by multi-scale foreground and background information. Similarly, the new context features are generated by the novel compact supervised codebook. Our compact supervised codebook minimizes irrelevant and redundant information, which, in result, achieves the lower-sized contextual feature vector. Finally, we combine both content and context features to represent the scene image. Our experiments on three widely used benchmark scene datasets using Support Vector Machine (SVM) classifier reveal that our proposed context and content features produce better results than existing context and content features, respectively. The fusion of the proposed two types of features significantly outperform numerous state-of-the-art features. • Propose a novel contextual feature extraction method for scene images. • Propose a novel unique tokens extraction algorithm. • Aggregate multi-scale content and contextual features to represent the scene images. • Evaluate our proposed method on three commonly-used benchmark datasets. [ABSTRACT FROM AUTHOR]

Band selection is a research hotspot in hyperspectral image processing. The continuity of the spectral bands causes the adjacent bands to be highly correlated, and correlation among long-range bands is possible with hundreds of spectral bands. Most existing deep learning methods fail to make full use of the inter-band correlation for band selection. In this paper, a novel dual-graph convolutional network based on band attention and a sparse constraint is proposed for band selection. The network consists of two branches. In the attention branch, band-based dual graphs are constructed to encode the contextual correlation of adjacent bands and the structural correlation of long-range bands into non-Euclidean space. Subsequently, the graph convolution-based band attention mechanism is devised to aggregate the band information in the band-based dual graphs and to generate the attention map for all bands. The band attention map is sparsely constrained and embedded as a mask into the trunk branch. In the trunk branch, sample-based dual graphs are constructed to represent the topological information of the samples in the spectral and spatial domains. Furthermore, a dense graph convolutional network is designed to extract and fuse the spatial–spectral and topological features from the shallow to deep layers for classification. A soft-shifting optimization strategy is implemented by defining a new loss from full bands and selected bands to solve the optimization problem caused by the sparse constraint. In this manner, band selection, feature extraction, and classification can be combined into an end-to-end trainable network. The experimental results on representative hyperspectral image datasets demonstrate the superiority of the proposed method over current state-of-the-art band selection methods. • An improved graph convolutional network is proposed for band selection. • Band-based and sample-based graphs are built with spatial and spectral information. • Design soft-shifting optimization strategy to overcome the optimization problem. • Mini batches of labeled and unlabeled samples are used for graph construction. [ABSTRACT FROM AUTHOR]

Accurately segmenting adjacent regions in images is one of the greatest challenges in image processing. In this paper, we introduce a double level set segmentation model based on mutual exclusion, which can accurately and independently segment adjacent regions. In order to avoid the creation of intersection regions and ensure the independence of neighboring regions, the model uses two level set functions, and defines the mutual exclusion term of neighboring regions according to the area of the joint region segmented by the two level set functions. Furthermore, we manually set the approximate contours of the target region as a prior information, defining prior constraints in the energy functional to ensure that the model only segments the organization we wanted. Combining the data term, the mutual exclusion term, the prior constraint term, the length term and the regularization term, we give the level set formulation of this double level sets energy functional and apply the gradient descent algorithm to solve its energy minimization problem. Applying the double level sets model to brain MR images, the experimental results show that the model can accurately and independently segment adjacent tissues in brain. The comparison experimental results with other models show that the double level sets model has a higher accuracy than the classical models in segmenting adjacent tissues. In addition, experiments on synthetic images show that the double level sets model can accurately segment images with intensity inhomogeneity and is robust to noise. Code is avaliable at https://github.com/ruicx/Mutual-Exclusion-Level-Set. [ABSTRACT FROM AUTHOR]

Infrared pedestrian segmentation is a challenging problem in infrared image processing. Kaniadakis entropy thresholding method can segment the infrared image with long tail distribution histogram. However, it failed in images with noise and complex background. In this paper, a novel infrared pedestrian segmentation algorithm based on the two-dimensional Kaniadakis entropy thresholding is proposed to address this problem. First, in order to introduce the spatial information of pixels, the two-dimensional Kaniadakis entropy thresholding algorithm is proposed through extending the one-dimensional Kaniadakis entropy thresholding algorithm and using the two-dimensional histogram of image. The recursive formulas of the two-dimensional algorithm are also presented to improve the computation efficiency. Second, an intensity suppressed strategy is embedded in the optimal thresholds searching process of the two-dimensional Kaniadakis entropy thresholding algorithm to segment the pedestrian in infrared images with complex background. In the experiment, the proposed algorithm is compared with the state-of-the-art image segmentation methods on several infrared images with different backgrounds. The experimental results verify the effectiveness of the proposed method both qualitatively and quantitatively. • Extend the Kaniadakis entropy thresholding to 2d. • Derive the recursive formulas of the 2d Kaniadakis entropy thresholding. • Propose a suppressed strategy in 2d-Kan for infrared pedestrian segmentation. • Verify the effectiveness of the proposed infrared pedestrian segmentation algorithm. [ABSTRACT FROM AUTHOR]

6D object pose estimation plays an important role in various applications such as robot manipulation and virtual reality. In this paper, we introduce a graph convolution neural network based method to addresses the problem of estimating the 6D pose of objects from a single RGB-D image. The proposed method fuses the appearance feature of the RGB image with the geometry feature of point clouds to predict pixel-level pose and the network also predicts pixel-level confidences to prune outlier predictions. The inner structure information of point cloud is learned by a graph convolution neural network. Specially, we adopt a residual graph convolution module to learn a discriminative feature. Our network enables end-to-end training and fast inference. The extensive experiments verify the method and the model achieves state-of-the-art for the LINEMOD and LINEMOD-OCCLUSION dataset (ADD-S: 88.68 and 65.38 respectively). [ABSTRACT FROM AUTHOR]

Image segmentation process is one of the most interesting and challenging problems in digital image processing tasks. The segmentation process involves finding similar regions within an image. Many segmentation problems are achieved by the incorporation of clustering techniques. One of the most common technique for clustering process is the Fuzzy C-means (FCM) algorithm. However, even when FCM is one of the most popular techniques applied in image segmentation, it presents some issues such as large computational time complexity, noise sensitivity, and initial cluster centers dependency. In order to solve these problems, this paper presents a Histogram Based Fuzzy Clustering (HBFC) technique using an improved version of Firefly Algorithm (FA). In the proposed approach, the FA involves three search strategies: rough set-based population, random attraction and local search mechanism. Also, the clustering process is conducted based on gray level histograms instead of single pixels of an image. Under such circumstances, the occurrence of misclassification of pixels is reduced. A rigorous comparative study is conducted among the proposed approach and several state-of-art Nature-Inspired Optimization Algorithms (NIOAs) and traditional clustering techniques. The numerical results indicate that the proposed approach outperform the well-known NIOA based clustering methods in terms of precision, robustness and quality of the segmented outputs. • An improved Firefly algorithm has been proposed for Image Clustering. • Random Attraction model and rough set have been utilized. • The proposed approach is a histogram based fast fuzzy clustering model. • The proposed method is compared against state-of-art segmentation methods. • Numerical results indicate the superior performance of the proposed method. [ABSTRACT FROM AUTHOR]

Zero-watermarking schemes are currently dedicated to resisting general desynchronization attacks, such as rotation and scaling, but cannot effectively resist complex desynchronization attacks, such as translation, cropping, length–width ratio changing, UnZign, and zero-padding. To this end, a robust zero-watermarking scheme resistant to complex desynchronization attacks was proposed in this paper based on local feature regions (LFRs) and quaternion polar harmonic Fourier moments (QPHFMs). First, the stable feature points of original color medical images were extracted with a speeded-up robust feature (SURF) operator; then, the stable LFRs were obtained through a feature scale-adaptive process and scale-standardized tests. Finally, the QPHFMs of the LFRs were calculated, and multiple feature sequences were constructed using their magnitudes and binarized to generate multiple zero-watermarks. The experimental results indicate that the proposed scheme can not only resist common image processing attacks and general desynchronization attacks but also robustly resist complex desynchronization attacks, such as translation, cropping, length–width ratio changing, UnZign, and zero-padding. The proposed scheme is superior to the state-of-the-art LFR-based multiple embedded watermarking schemes and zero-watermarking schemes. • This scheme is the first multiple color medical image zero-watermarking scheme. • It constructs multiple zero-watermarks and effectively improves the reliability. • It does not change the original image, and ensures image integrity. • It can resist complex desynchronization attacks such as cropping and Unzign, etc. • It has better performance than state-of-the-art multiple watermarking schemes. [ABSTRACT FROM AUTHOR]

Image Contrast Enhancement (ICE) is a crucial step in several image processing and computer vision applications. Its main objective is to improve the quality of the visual information contained in the processed images. The presence of noise and small sets of pixels in images are not only irrelevant for their visualization. It also negatively affects the improvement process of ICE schemes since the inclusion of irrelevant information avoids the appropriate distribution of significant pixel intensities in the enhanced image. As a consequence of this effect, most of the proposed ICE methods present different associated problems such as the production of undesirable artifacts, noise amplification, over saturation and bad human visual perception. In this paper, an Image Contrast Enhancement (ICE) method for grayscale and color images is presented. The proposed approach has the propriety of eliminating noisy and irrelevant information in order to improve the distribution capacity of significant pixel intensities in the enhanced image. Our method eliminates multiple groups of a very small number of pixels that, according to their characteristics, do not represents any object or important detail of the image. This process is done by the Mean-shift algorithm, which is used to replace such sets of irrelevant pixels in the original histogram by significant pixel densities represented by local maxima. Then, the Moth Swarm Algorithm (MSA) is used to redistribute the pixel intensities of the reduced histogram so that the value from Kullback–Leibler entropy (KL-entropy) has been maximized. The proposed approach has been tested considering different public datasets commonly used in the literature. Its results are also compared with those produced by other well-known ICE techniques. Evaluation of the experimental results demonstrates that the proposed approach highlights the important details of the image also improving its human visual appearance. • It uses an elimination process to reduce irrelevant features. • It uses the KL-entropy for the enhancement process. • It considers the use of the MSA to find the best enhanced image. • Its results are competitive in comparison with other state-of-the art methods. [ABSTRACT FROM AUTHOR]

Breast cancer is the most commonly diagnosed cancer among women. Convolutional neural networks (CNN)-based mammogram classification plays a vital role in early breast cancer detection. However, it pays too much attention to the lesions of mammograms and ignores the global characteristics of the breast. In the process of diagnosis, doctors not only pay attention to the features of local lesions but also combine with the comparison to the global characteristics of breasts. Mammogram images have a visible characteristic, which is that the original image is large, while the lesions are relatively small. It means that the lesions are easy to overlook. This paper proposes an automated multi-scale end-to-end deep neural networks model for mammogram classification, that only requires mammogram images and class labels (without ROI annotations). The proposed model generated three scales of feature maps that make the classifier combine global information with the local lesions for classification. Moreover, the images processed by our method contain fewer non-breast pixels and retain the small lesions information as much as possible, which is helpful for the model to focus on the small lesions. The performance of our method is verified on the INbreast dataset. Compared to other state-of-the-art mammogram classification algorithms, our model performs the best. Moreover, the multi-scale method is applied to the networks with fewer parameters that can achieve comparable performance, while saving 60% of the computing resources. It shows that the multi-scale method can work for both performance and computational efficiency. [ABSTRACT FROM AUTHOR]

In this paper, we propose a new subspace clustering method based on alignment and graph embedding (SCAGE). In SCAGE, we unify the image alignment process and clustering subspace learning process based on low rank and sparse representation. Besides, we use the label prediction information, error information and coefficients to conduct the graph embedding. In addition, the prior knowledge is used to initialize the label prediction matrix which not only speeds up the converge of the clustering process but also achieves a better result. Various experiments show that SCAGE achieves better performance than state-of-the-art algorithms. [ABSTRACT FROM AUTHOR]