Deep learning models have been widely applied to Polarimetric Synthetic Aperture Radar (PolSAR) image classification due to their excellent performance. However, unlike natural images, PolSAR data is a 3×3 covariance matrix for each resolution unit. Existing deep learning methods generally convert the covariance matrix into a vector as the input of neural networks, which destroys the correlation between channels and distorts the matrix structure. To alleviate this issue, we explore a Symmetric Positive Definite (SPD) convolution network for PolSAR images, which directly inputs the PolSAR complex matrix into the network to learn the geometric features in Riemannian space. Furthermore, a CNN-enhanced SPDnet is designed to further learn the contextual high-level features, which can convert Riemannian matrix features into Euclidean space and apply them for classification. Experimental results on real PolSAR data sets demonstrate the proposed method can achieve better performance than the state-of-the-art methods.

Laryngeal leukoplakia classification is challenging using white light endoscopy images. Relevant research focus on normal tissues versus non normal tissues, cancer versus non cancer classification. The objective of this paper is to classify laryngeal leukoplakia in white light endoscopy images into six classes: normal tissues, inflammatory keratosis, mild dysplasia, moderate dysplasia, severe dysplasia and squamous cell carcinoma. We proposed a dense multiscale convolutional neural network including parallel multiscale convolution, dense convolution and recurrent convolution in favor of extracting dense multiscale features of laryngeal leukoplakia for fine classification. The proposed network achieved an overall accuracy of 0.8958 for the six-class classification. It has high sensitivity and specificity for each class which are, respectively, 1.0000 and 0.9394 for normal tissues, 0.6667 and 1.0000 for inflammatory keratosis, 0.8889 and 0.9744 for mild dysplasia and moderate dysplasia, 0.7500 and 1.0000 for severe dysplasia, 1.0000 and 0.9767 for squamous cell carcinoma. The experimental results show that our proposed model is superior to the state-of-the-art deep learning-based models.

Abstract Deraining has been attracting a lot of attention from researchers, and various methods have been proposed, especially deep‐networks are widely adopted in recent years. Their structures and learning become more and more complicated and diverse, making it difficult to analyze the contributions and improvements. In this paper, a comprehensive review for current rain removal methods is first provided to show their contributions. Specifically, they are reviewed in terms of handing rain streaks and rain mist. Second, besides evaluating their rain removal ability, they are also evaluated in terms of their impact on subsequent stereo‐matching task. To this end, a new deraining dataset is first prepared, called Rain‐Kitti2012 and Rain‐Kitti2015. They are created by adding rain part to clean image‐pairs in Kitti2012 and Kitti2015. By then, nine state‐of‐the‐art deraining methods are evaluated with full‐reference and no‐reference image quality assessment metrics. Furthermore, the blurriness and distortion types introduced during deraining are measured. Finally, three learning‐based stereo matching methods are compared, and they take the outputs of deraining methods as inputs. It is further discussed how derained images influence the accuracy of stereo matching, which can provide some insight for jointly handling rain removal and stereo matching. 1: A comprehensive review for the current rain removal methods is provided. They are categorized into rain‐streak‐oriented and rain‐mist‐oriented approaches in terms of degradation type, and are categorized into model‐driven and data‐driven approaches in terms of methodology. 2: A new image deraining dataset is introduced, which is the first dataset that can be used to perform stereo‐matching‐driven evaluation for deraining methods. The dataset is created by adding rain part to clean images in KITTI2012 and KITTI2015. 3: We evaluate 9 deep learning based deraining methods with full‐reference and no‐ reference metrics. In addition, the types of distortions produced by these methods are discussed and measured quantitatively. And, the impact of 9 deraining methods on the subsequent stereo matching task is evaluated, which can provide some insight on how to design stereo matching task‐driven deraining methods.

Abstract This article presents a saliency guided remote sensing image dehazing network model. It consists of the following three blocks: A dense residual based backbone network, a saliency map generator, and a deformed atmospheric scattering model (ASM) based haze removal model, of which the dense residual based backbone network is used to capture the texture detail information of a remote sensing image, the saliency map generator is used to generate the saliency map of the related remote sensing image, and the generated saliency map is used to guide the network to capture more texture details through the guided fusion module. Finally, the deformed atmospheric scattering model (ASM) is used to remove haze from remote sensing images. The model here is compared with several state‐of‐art dehazing methods on synthetic data sets and real remote sensing images. Experimental results show that on the synthetic data set, the PSNR value of this model is increased by 4.47 db and the SSIM value is increased by 0.045 compared with the best model. On real remote sensing hazy images, the visual effect of our model is also better than that of existing methods. The authors also perform experiments to demonstrate that remote sensing image dehazing is helpful for remote sensing image detection automatically.

Abstract Underwater images usually suffer from colour distortion, blur, and low contrast, which hinder the subsequent processing of underwater information. To address these problems, this paper proposes a novel approach for single underwater images enhancement by integrating data‐driven deep learning and hand‐crafted image enhancement techniques. First, a statistical analysis is made on the average deviation of each channel of input underwater images to that of its corresponding ground truths, and it is found that both the red channel and the green channel of an underwater image contribute to its colour distortion. Concretely, the red channel of an underwater image is usually seriously attenuated, and the green channel is usually over strengthened. Motivated by such an observation, an attention mechanism guided residual module for underwater image colour correction is proposed, where the colour of the red channel of the underwater image and that of the green channel is compensated in a different way, respectively. Coupled with an attention mechanism, the residual module can adaptively extract and integrate the most discriminative features for colour correction. For scene contrast enhancement and scene deblurring, the traditional image enhancement techniques such as CLAHE (contrast limited adaptive histogram equalization) and Gamma correction are coupled with a multi‐scale convolutional neural network (MSCNN), where CLAHE and Gamma correction are used as complement to deal with the complex and changeable underwater imaging environment. Experiments on synthetic and real underwater images demonstrate that the proposed method performs favourably against the state‐of‐the‐art underwater image enhancement methods.

Neuron detection is a key step in individualizing and counting neurons which are important for assessing physiological and pathophysiological information. A large number of methods including deep learning networks have been proposed but mainly targeting regions with few aggregated neurons. The objective of this paper is to address an automated neuron detection problem in heterogeneous hippocampus region with different degrees of neuron aggregation. Since deep learning networks require a lot of ground truths but neuron instance annotation is impossible in regions where numerous neurons are clustered, ground truth of centroids marked at the center of neurons is created for training. We propose a multiscale convolutional neural network (CNN) to regress neuron centroid mapping across image. Using multiscale information makes the proposed network applicable not only for single individual neurons, but also for a large number of aggregated neurons. Experimental results show that our method is superior to state-of-the-art deep learning-based algorithms. To our knowledge, this is the first deep learning study to detect neurons in regions of highly clustered neurons.

Images captured in the sand–dust weather often suffer from serious colour cast and poor contrast, and this has serious implications for outdoor computer vision systems. To address these problems, a normalised gamma transformation‐based contrast‐limited adaptive histogram equalisation (CLAHE) with colour correction in Lab colour space for sand–dust image enhancement is proposed in this study. This method consists of image contrast enhancement and image colour correction. To avoid producing new colour deviation, the input sand–dust images are first transformed from red, green, and blue colour space into Lab colour space. Then, the contrast of the lightness component (L channel) of the sand–dust image is enhanced using CLAHE. To avoid unbalanced contrast, as well as to reduce the overincreased brightness caused by CLAHE, a normalised gamma correction function is introduced to CLAHE. After that, the a and b chromatic components are recovered by a grey‐world‐based colour correction method. Experiments on real sand–dust images demonstrate that the proposed method can obtain the highest percentage of new visible edges for all testing images. The contrast restoration exhibits good colour fidelity and proper brightness.

Image haze removal is highly desired for the application of computer vision. This study proposes a novel context‐guided generative adversarial network (CGGAN) for single image dehazing. Of which, a novel new encoder–decoder is employed as the generator. In addition, it consists of a feature‐extraction net, a context‐extraction net, and a fusion net in sequence. The feature‐extraction net acts as an encoder, and is used for extracting haze features. The content‐extraction net is a multi‐scale parallel pyramid decoder and is used for extracting the deep features of the encoder and generating coarse dehazing image. The fusion net is a decoder and is used for obtaining the final haze‐free image. In order to get better dehazing results, multi‐scale information obtained during the decoding process of the context extraction decoder is used for guiding the fusion decoder. By introducing an extra coarse decoder to the original encoder–decoder, the CGGAN can make better use of the deep feature information extracted by the encoder. To ensure that the proposed CGGAN works effectively for different haze scenarios, different loss functions are employed for the two decoders. Experiments results show the advantage and the effectiveness of the proposed CGGAN, evidential improvements over existing state‐of‐the‐art methods are obtained.

Rain image enhancement is important for outdoor computer vision applications. In this study, the authors propose a multi‐stage filtering method for single rainy image enhancement. It is based on their new rainy image model, and consists of two main operations: rain streaks removal and rain fog removal. For rain streaks removal, based on one key observation that the low‐pass version of a rainy image and that of a non‐rainy image of the same scene are almost the same after appropriate low‐pass filtering, they remove rain streaks from rainy images by decomposing an input rainy image (or a rainy component image) into the low‐frequency (LF) part and the high‐frequency (HF) part via an LF smooth filter, i.e. the traditional Gaussian filter with a simple subtraction operation in multiple different stages. After rain streaks removal, dark channel prior‐based method was employed for rain fog removal. Experimental results show that the proposed algorithm generated comparable outputs with most of the state‐of‐the‐art algorithms with low computation cost.

Images captured in hazy weather are usually of poor quality, which has a negative effect on the performance of outdoor computer imaging systems. Therefore, haze removal is critical for outdoor imaging applications. In this paper, a quick single-image dehazing method based on a new effective image prior, luminance dark prior, was proposed. This new image prior arose from the observation that most local patches in the luminance image of a haze-free outdoor YUV color space image usually contain pixels of very low intensity, which is similar to the dark channel prior used with HE for RGB images. Using this new prior, a transmission map was used to estimate the thickness of the haze in an image directly from the luminance component of the YUV color image. To obtain a transmission map with a clear edge outline and depth layer of scene objects, a joint filter containing a bilateral filter and Laplacian operator was employed. Experimental results demonstrated that the proposed method unveiled details and recovered vivid colors even in heavily hazy regions, and provided superior visual effects to many other existing methods.

Image noise can severely affect Chinese tablet image comprehension. In this paper, a novel integrated denoising and enhancement method for Chinese tablet image is proposed. The method consists of three stages de-noising operations. First, granular bright spots are smoothed by convoluting input Chinese tablet images with the bilateral filter. For obtaining more clear edge detail image, contrast enhancement between the foreground and background of the Chinese tablet image is followed by Top-Hat and Bottom-Hat transformations. Next, a mixture of run-length statistics and connected region techniques is employed to remove the random block noises in the images. Then, two mathematical morphological operators, erosion and dilation, are used to remove small holes and linear noises. Experimental results show that the proposed method can effectively remove most image noise (including block noise, and linear noise) and preserve characters better than existing methods.

Computer-aided diagnostic (CAD) schemes often employ a filter for enhancement of lesions as a preprocessing step for improving sensitivity and specificity. The filter enhances objects similar to a model employed in the filter; e.g., a blob enhancement filter based on the Hessian matrix enhances sphere-like objects. Actual lesions, however, often differ from a simple model, e.g., a lung nodule is generally modeled as a solid sphere, but there are nodules of various shapes and with inhomogeneities inside such as a spiculated one and a ground-glass opacity. Thus, conventional filters often fail to enhance actual lesions. Our purpose in this study was to develop a supervised filter for enhancement of lesions by use of a massive-training artificial neural network (MTANN) in a computer-aided diagnostic (CAD) scheme for detection of lung nodules in CT. The MTANN filter was trained with actual nodules in CT images to enhance actual patterns of nodules. By use of the MTANN filter, the sensitivity and specificity of our CAD scheme were improved substantially. With the database with 69 lung cancers, our CAD scheme with the MTANN filter achieve a 97% sensitivity with 6.7 false positives (FPs) per section, whereas a conventional CAD scheme with a difference-image technique achieved a 96% sensitivity with 19.3 FPs per section.

A general relation between the cumulant functions and the direction of arrival (DOA) parameters is developed. Using this relation it is shown that several DOA estimation algorithms can be obtained. One of these algorithms, involving the use of a Hankel matrix, is developed and its performance is studied in computer simulations.>

The harmonic retrieval problem considered concerns the estimation of the frequencies in a sum of complex sinusoids in the presence of Gaussian measurement noise. The approach used involves representing this problem as a realization problem in the state space. Fourth-order cumulants are used as a signal processing tool. It is shown that the estimation of the unknown frequencies can be achieved by solving a generalized eigenvalue problem involving a Hankel matrix composed from the fourth-order cumulants of the measured output. Simulation results show the effectiveness of this method when the signals are corrupted by Gaussian noises (white or colored).>