Search

Low-light image enhancement is a crucial preprocessing task in complex vision tasks. It directly impacts object detection, image segmentation, and image recognition outcomes. In recent years, with the continuous development of deep learning techniques, an increasing number of image enhancement methods based on deep learning have emerged. However, due to the high cost of data collection and the limited content of supervised learning datasets, more and more scholars have shifted their focus to the field of unsupervised image enhancement. Unsupervised image enhancement methods do not require paired images of the same scene during the training process, which greatly reduces the threshold for network training. Nevertheless, current unsupervised methods still suffer from issues such as unstable enhancement effects and limited generalization ability. To address these problems, we propose an improved low-light image enhancement method. The proposed method employs the LSGAN as the training architecture and utilizes an attention map network to dynamically generate attention maps that best fit the network enhancement task, which can effectively improve the generalization ability and enhancement performance of the network. Additionally, we adopt an attention mechanism to enhance the subtle details of the image features. Regarding the network training, considering that the traditional convolutional neural network discriminator may not provide effective guidance to the generator in the early stages of training, we propose an improved discriminator structure. The experimental results demonstrate that our method can achieve good enhancement performance on different datasets and has practical value. Although our method has advantages in enhancing low-light images, it also has certain limitations, such as the network size not meeting the requirements for lightweight models and the potential for further improvement under extremely low-light conditions. We will strive to address these issues as comprehensively as possible in our future research.

The research on hyperspectral anomaly detection algorithms has become a hotspot, driven by a lot of practical applications, such as mineral exploration, environmental monitoring and the national defense force. However, most existing hyperspectral anomaly detectors are designed with a single pixel as unit, which may not make full use of the spatial and spectral information in the hyperspectral image to detect anomalies. In this paper, to fully combine and utilize the spatial and spectral information of hyperspectral images, we propose a novel spectral-based selective searching method for hyperspectral anomaly detection, which firstly combines adjacent pixels with the same spectral characteristics into regions with adaptive shape and size and then treats those regions as one processing unit. Then, by fusing adjacent regions with similar spectral characteristics, the anomaly can be successfully distinguished from background. Two standard hyperspectral datasets are introduced to verify the feasibility and effectiveness of the proposed method. The detection performance is depicted by intuitive detection images, receiver operating characteristic curves and area under curve values. Comparing the results of the proposed method with five popular and state-of-the-art methods proves that the spectral-based selective searching method is an accurate and effective method to detect anomalies.

Numerous applications are hindered by shadows in high resolution satellite remote sensing images, like image classification, target recognition and change detection. In order to improve remote sensing image utilization, significant importance appears for restoring surface feature information under shadow regions. Problems inevitably occur for current shadow compensation methods in processing high resolution multispectral satellite remote sensing images, such as color distortion of compensated shadow and interference of non-shadow. In this study, to further settle these problems, we analyzed the surface irradiance of both shadow and non-shadow areas based on a satellite sensor imaging mechanism and radiative transfer theory, and finally develop an irradiance restoration based (IRB) shadow compensation approach under the assumption that the shadow area owns the same irradiance to the nearby non-shadow area containing the same type features. To validate the performance of the proposed IRB approach for shadow compensation, we tested numerous images of WorldView-2 and WorldView-3 acquired at different sites and times. We particularly evaluated the shadow compensation performance of the proposed IRB approach by qualitative visual sense comparison and quantitative assessment with two WorldView-3 test images of Tripoli, Libya. The resulting images automatically produced by our IRB method deliver a good visual sense and relatively low relative root mean square error (rRMSE) values. Experimental results show that the proposed IRB shadow compensation approach can not only compensate information of surface features in shadow areas both effectively and automatically, but can also well preserve information of objects in non-shadow regions for high resolution multispectral satellite remote sensing images.

Conventionally, high dynamic-range (HDR) imaging is based on taking two or more pictures of the same scene with different exposure. However, due to a high-speed relative motion between the camera and the scene, it is hard for this technique to be applied to push-broom remote sensing cameras. For the sake of HDR imaging in push-broom remote sensing applications, the present paper proposes an innovative method which can generate HDR images without redundant image sensors or optical components. Specifically, this paper adopts an area array CMOS (complementary metal oxide semiconductor) with the digital domain time-delay-integration (DTDI) technology for imaging, instead of adopting more than one row of image sensors, thereby taking more than one picture with different exposure. And then a new HDR image by fusing two original images with a simple algorithm can be achieved. By conducting the experiment, the dynamic range (DR) of the image increases by 26.02 dB. The proposed method is proved to be effective and has potential in other imaging applications where there is a relative motion between the cameras and scenes.

Shadow often results in difficulties for subsequent image applications of multispectral satellite remote sensing images, like object recognition and change detection. With continuous improvement in both spatial and spectral resolutions of satellite remote sensing images, a more serious impact occurs on satellite remote sensing image interpretation due to the existence of shadow. Though various shadow detection methods have been developed, problems of both shadow omission and nonshadow misclassification still exist for detecting shadow well in high-resolution multispectral satellite remote sensing images. These shadow detection problems mainly include high small shadow omission and typical nonshadow misclassification (like bluish and greenish nonshadow misclassification, and large dark nonshadow misclassification). For further resolving these problems, a new shadow index is developed based on the analysis of the property difference between shadow and the corresponding nonshadow with several multispectral band components (i.e., near-infrared, red, green and blue components) and hue and intensity components in various invariant color spaces (i.e., HIS, HSV, CIELCh, YCbCr and YIQ), respectively. The shadow mask is further acquired by applying an optimal threshold determined automatically on the shadow index image. The final shadow image is further optimized with a definite morphological operation of opening and closing. The proposed algorithm is verified with many images from WorldView-3 and WorldView-2 acquired at different times and sites. The proposed algorithm performance is particularly evaluated by qualitative visual sense comparison and quantitative assessment of shadow detection results in comparative experiments with two WorldView-3 test images of Tripoli, Libya. Both the better visual sense and the higher overall accuracy (over 92% for the test image Tripoli-1 and approximately 91% for the test image Tripoli-2) of the experimental results together deliver the excellent performance and robustness of the proposed shadow detection approach for shadow detection of high-resolution multispectral satellite remote sensing images. The proposed shadow detection approach is promised to further alleviate typical shadow detection problems of high small shadow omission and typical nonshadow misclassification for high-resolution multispectral satellite remote sensing images.

Shadows in very high-resolution multispectral remote sensing images hinder many applications, such as change detection, target recognition, and image classification. Though a wide variety of significant research has explored shadow detection, shadow pixels are still more or less omitted and are wrongly confused with vegetation pixels in some cases. In this study, to further manage the problems of shadow omission and vegetation misclassification, a mixed property-based shadow index is developed for detecting shadows in very high-resolution multispectral remote sensing images based on the difference of the hue component and the intensity component between shadows and nonshadows, and the difference of the reflectivity of the red band and the near infrared band between shadows and vegetation cover in nonshadows. Then, the final shadow mask is achieved, with an optimal threshold automatically obtained from the index image histogram. To validate the effectiveness of our approach for shadow detection, three test images are selected from the multispectral WorldView-3 images of Rio de Janeiro, Brazil, and are tested with our method. When compared with other investigated standard shadow detection methods, the resulting images produced by our method deliver a higher average overall accuracy (95.02%) and a better visual sense. The highly accurate data show the efficacy and stability of the proposed approach in appropriately detecting shadows and correctly classifying shadow pixels against the vegetation pixels for very high-resolution multispectral remote sensing images.

The digital time delay integration (digital TDI) technology of the complementary metal-oxide-semiconductor (CMOS) image sensor has been widely adopted and developed in the optical remote sensing field. However, the details of targets that have low illumination or low contrast in scenarios of high contrast are often drowned out because of the superposition of multi-stage images in digital domain multiplies the read noise and the dark noise, thus limiting the imaging dynamic range. Through an in-depth analysis of the information transfer model of digital TDI, this paper attempts to explore effective ways to overcome this issue. Based on the evaluation and analysis of multi-stage images, the entropy-maximized adaptive histogram equalization (EMAHE) algorithm is proposed to improve the ability of images to express the details of dark or low-contrast targets. Furthermore, in this paper, an image fusion method is utilized based on gradient pyramid decomposition and entropy weighting of different TDI stage images, which can improve the detection ability of the digital TDI CMOS for complex scenes with high contrast, and obtain images that are suitable for recognition by the human eye. The experimental results show that the proposed methods can effectively improve the high-dynamic-range imaging (HDRI) capability of the digital TDI CMOS. The obtained images have greater entropy and average gradients.

Autofocus methods are conventionally based on capturing the same scene from a series of positions of the focal plane. As a result, it has been difficult to apply this technique to scanning remote sensing cameras where the scenes change continuously. In order to realize autofocus in scanning remote sensing cameras, a novel autofocus method is investigated in this paper. Instead of introducing additional mechanisms or optics, the overlapped pixels of the adjacent CCD sensors on the focal plane are employed. Two images, corresponding to the same scene on the ground, can be captured at different times. Further, one step of focusing is done during the time interval, so that the two images can be obtained at different focal plane positions. Subsequently, the direction of the next step of focusing is calculated based on the two images. The analysis shows that the method investigated operates without restriction of the time consumption of the algorithm and realizes a total projection for general focus measures and algorithms from digital still cameras to scanning remote sensing cameras. The experiment results show that the proposed method is applicable to the entire focus measure family, and the error ratio is, on average, no more than 0.2% and drops to 0% by reliability improvement, which is lower than that of prevalent approaches (12%). The proposed method is demonstrated to be effective and has potential in other scanning imaging applications.

It is difficult to deal with the auto-focus of aerial push-broom remote sensing camera (APRC) based on image processing. First of all, this paper recommended the Splicing structure of CCD in the APRC Based on this, auto-focus method which is made use of the overlapping area of CCD mosaic structure was proposed, combined with the characteristic of the imaging mode. After analyzing of experiments about all kinds of focus-examine functions based on image processing, the gradient square function has been chosen. The experimental results show that the proposed auto-focus method for the APRC is proved to be feasible and real-time.

For multi-channel high-speed image data of certain space remote sensing camera, this paper proposed a new image data acquisition and storage system. First of all, the system uses Field-Programmable Gate Array(FPGA) controlling synchronous dynamic random access memory(SDRAM) array to realize the cache of image data, then according to the received commands, sends certain channel's image data to the acquisition card to write the image data into SCSI hard disk and implements the final storage. A Time Delay Integral Charge Couple Device(TDICCD) image data acquisition and storage system with five channels is developed, pixel clock rate is up to 96MHz,total effective data rate is up to 2.88Gb/s，single channel storage depth is up to 7810 lines. The experimental results indicate that this system work stably, high effectively which can meet the requirement of acquisition and storage of multi-channel high-speed TDICCD image data.

The volume of raw image data captured by the high resolution camera is extremely huge. Thus the efficient image compression method should be used to decrease the bit rate. The image compression method based on wavelet is used more widely nowadays. However, two dimensional wavelet is only the tensor product of the one dimensional wavelet whose support region of basis function is extended from interval to square. Contourlet is an image multiscale geometric analysis tool, which could represent image sparsely and has strong capability of nonlinear approximation. The basis function of contourlet is multi-directional and anisotropic. Nevertheless, contourlet is redundant. So the non-redundant Wavelet Based Contourlet Transform (WBCT) is used in this paper. The SPIHT algorithm is very efficient way to coding the significant coefficients. And the improved no lists SPIHT is more easy to implemented by hardware. Image compression method based on the combination of both wavelet based contourlet transform and no lists SPIHT coding is proposed in the paper. Experiment shows that compared to wavelet based scheme the contourlet scheme can reserve the texture of the image. For barbara test image when coding at low bit rate the PSNR can improve about 0.2dB.

In order to identify the vibration parameters of a vibrated image, the principle and shortage of the traditional methods that based on image derivative and image transformation are analyzed, and a new parameter identification method is proposed afterward. By placing a high-speed CCD aside the master CCD, the vibration can be sampled accurately, and more information can be obtained for parameter identification. The series frames that captured by high-speed CCD are processed by gray projection algorithm and the vibration extents between adjacent frames along x and y direction are obtained. By using simple triangle operations, the vibration parameters can be identified, and the computational complexity of which can be reduced greatly. The experiment shows that the parameter identification method based on high-speed CCD can identify the parameters of vibration with any form accurately. The precise of parameter identification for linear vibration is less than 2.95%.

Pre-amplifier circuit is critical for the noise performance of the high speed CCD camera. Its main functions are amplification and impedance transform. The high speed and low noise pre-amplifier of CCD camera is discussed and designed in this paper. The high speed and low noise operational amplifier OPA842 is adopted as the main part. The gain-set resistors for the amplifier are designed optimally. The different precision gain-set resistors are swept using Monte Carlo method. CCD video signal which has high DC offset voltage is AC coupled to the amplifier. The output signal of the amplifier is source terminated using 50 ohms matching resistor so as to transmit the video signal through coaxial cable. When the circuit works in high speed, the PCB will have important effect to circuit performance and can even cause the amplifier unstable due to the parasitic problem of PCB. So the parasitic model of the PCB is established and the PCB layout design issues are also presented. The design result shows that the pre-amplifier can be used in the camera whose pixel rate could be up to 40 MHz and its input referred noise density is about 3 nV/Hz1/2.