Search

A modified multiway pixel-value differencing method for image steganography using general quantization ranges of pixel pairs’ difference values is proposed, where the widths of the quantization ranges are not limited to be powers of two. The method can be employed to embed a secret message into a cover image with a higher embedding rate. The cover image is partitioned into non-overlapping complete or incomplete blocks using $1 \times 3$ , $2 \times 2$ , $2 \times 3$ , or $3 \times 3$ block templates. The message bitstream, whose bit positions are randomized in advance by use of a random number generator, can be embedded in the resulting blocks. More specifically, one pixel in each block is set as a shared pixel, and the least-significant-bit substitution method with the optimal pixel adjustment process is employed to embed message bits into the shared pixel. Next, the shared pixel in the block is combined with each remaining pixel to form a pixel pair, which is then utilized to embed message bits by a new pixel-value differencing method. Also, the use of non-power-of-two range widths is accomplished by a multiple-based number conversion mechanism, and the shared pixel in each block, as well as each related pixel pair whose two grayscale values pass a falling-off-boundary check, are used to embed the digits of the resulting multiple-based number. The experimental results show that the embedding rates yielded by the proposed method are higher than those yielded by existing multiway pixel-value differencing methods. The stego-images resulting from embedding secret messages retain good image quality, and the RS steganalysis process cannot detect the presence of the embedded secret messages.

Latent low-rank representation has been applied to multi-level image decomposition for the fusion of infrared and visible images to obtain good results. However, when the original infrared and visible images are of low quality, the visual effect of the fused images is still unsatisfactory. To combat this challenge, this paper proposes an infrared and visible image fusion method based on multi-level latent low-rank representation joint with image enhancement and multiple visual weight information. First, the source images are decomposed into detail parts - including detail images and detail matrices - and the base images respectively using multi-level latent low-rank representation. Then the nuclear norm based fusion strategy is used to fuse the detail matrices and multi-visual weights determined by the clarity, local contrast and edge-corner saliency is used to fuse the detail images. The aforementioned two fusion results are weight averaged to obtain a fused detail image. The base images are fused by an averaging strategy after Retinex-based enhancement. The final fused image is obtained by combining the fused detail image and the fused base image. Compared with other state-of-the-art fusion methods, the proposed algorithm displays better fusion performance in both subjective and objective evaluation.

Principal Component Analysis (PCA) has been successfully applied to many applications, including ear recognition. This paper presents a 2D Wavelet based Multi-Band Principal Component Analysis (2D-WMBPCA) ear recognition method, inspired by PCA based techniques for multispectral and hyperspectral images. The proposed 2D-WMBPCA method performs a 2D non-decimated wavelet transform on the input image, dividing it into its wavelet subbands. Each resulting subband is then divided into a number of frames based on its coefficient’s values. The multi frame generation boundaries are calculated using either equal size or greedy hill climbing techniques. Conventional PCA is applied on each subband’s resulting frames, yielding its eigenvectors, which are used for matching. The intersection of the energy of the eigenvectors and the total number of features for each subband shows the number of bands which yield the highest matching performance. Experimental results on the images of two benchmark ear datasets, called IITD II and USTB I, demonstrated that the proposed 2D-WMBPCA technique significantly outperforms Single Image PCA by up to 56.79% and the eigenfaces technique by up to 20.37% with respect to matching accuracy. Furthermore, the proposed technique achieves very competitive results to those of learning based techniques at a fraction of their computational time and without needing to be trained.

This paper focuses on the unsupervised segmentation of images, which is an essential topic in the field of computer vision. In the absence of prior knowledge, it is challenging to generate semantic segmentation regions based on image content automatically. In this paper, we consider unsupervised image segmentation from the perspective of sub-region clustering and graph convolution. We over-segment the source image into disjoint sub-regions and generate multiscale representative maps for each sub-region. To explore the potential contextual correlation between different sub-regions, we build a graph model that establishes the dependency relationship and use a graph convolution network to transfer long-range contextual information. The segmentation results are obtained by combining sub-regions clustering and graph convolution training. We conduct extensive experiments on three image datasets, and the results show that the proposed method can provide consistent and meaningful segmentation results.

Sparse depth completion generates a dense depth image from its sparse measurement with the guidance of RGB image. In this paper, we propose attention guided sparse depth completion using convolutional neural networks, called AGNet. We adopt attention learning to get geometric cues for depth regression from RGB image and capture multi-scale depth structures. First, we use RGB image and valid binary mask from the input sparse depth image as input to generate an initial coarse depth image and its confidence map. Then, we generate attention map for depth refinement using a cross spatial attention module (CSAM). CSAM separately takes the RGB image with valid mask and invalid mask as input to make full use of color information in attention map. Next, we build a multi-scale learning network to encode the sparse depth image with different scales, thus leading to accurate depth completion. AGNet takes advantage of the input sparse depth image for encoding coarse features with a moderate model size. Experimental results show that AGNet achieves comparable performance with state-of-the-art methods for depth completion in NYU v2 dataset.

The feature of leaky cable fixture extracted by Local Binary Pattern (LBP) and its variants in high-speed railway tunnel has the defects of lacking description and high dimension. This paper proposes a new operator named Multi-scale Continuous Gradient Local Binary Pattern (MCG-LBP), which can realize the scale transformation of feature maps and ensure the low dimensionality of descriptors. For MCG-LBP, firstly a bi-directional triplet around the central pixel is presented to indicate the specific direction of gradient in circle neighborhood. Then, an effective dimensionality reduction strategy is introduced to perform successive down-sampling iterations. Finally, the multi-scale joint descriptors are encoded by continuous gradient sequences from different down-sampling maps, and Support Vector Machines is used to classify faulty cable fixtures. The proposed MCG-LBP can elicit a discriminative description through complementary gradient information generated by the combination of different single-scale features. While the low dimensionality of descriptor and no complex parameter to deal with both make it has higher computational efficiency. Experimental results show that the Recall and Precision of MCG-LBP reach 92.6% and 83.5% respectively on cable fixture data set, which is superior to the state-of-the-art methods.

The ground penetrating radar (GPR) data in the complex detection environment is non-stationary, non-Gaussian, and non-uniform, so the traditional noise attenuation methods are difficult to meet the requirements of denoising. Therefore, we introduced the K-singular value decomposition (K-SVD) dictionary learning into the denoising of GPR signals. It uses the orthogonal matching pursuit (OMP) algorithm to sparse decompose different radar data and trains the overcomplete dictionary with sample characteristics. K-SVD makes full use of the prior information and can extract features according to the sample data adaptively, which means it has strong sparse representation competence. Because radar signals can be sparsely represented in the dictionary, whereas the random noise does not have a sparse representation, the K-SVD dictionary can be used to distinguish effective signals from noise in the GPR data. We used the discrete cosine transform (DCT) and K-SVD dictionaries to process the Gaussian noise and stochastic clutter in the GPR profile. The results show that both K-SVD and DCT can effectively suppress the Gaussian noise. But for the clutter generated by the random medium, the DCT dictionary also causes damage to the effective signals while removing the noise; whereas the K-SVD dictionary learning algorithm uses the DCT dictionary as the initial dictionary and carries out adaptive learning on the noisy data, considering the information in the block and the global observation to complete the denoising, with good denoising effect and high fidelity. We finally verified the effectiveness and practicability of the K-SVD method for measured data.

The fade-in and fade-out algorithm based on the Bernstein polynomial has certain limitations in image fusion. Therefore, this article proposes a new image fusion algorithm. First, the SIFT algorithm is used to register the images. Second, for the disjointed case of overlapping regions, a progressive image mosaic fusion algorithm in the form of a sine function is proposed. Finally, in order to make the progressive image mosaic fusion algorithm suitable for a variety of overlapping regions, this paper adds segmentation technology. The simulation experiment results show that the algorithm proposed in this paper is in good agreement with the spatial details and texture details of a high-resolution panchromatic image, and the time is shorter, which meets the real-time requirements. In addition, the algorithm proposed in this paper is effective in applications such as virtual reality and art image fusion.

This paper proposes a basic taxonomy of image contours. Our goal is to classify smooth curves into five categories, namely, circles, ellipses, line segments, arcs of circles and arcs of ellipses. These geometrical structures have been chosen as they serve as input of many computer vision tasks. The proposed strategy is applied on a set of initial disjoint contours, which are grouped together to form the aforementioned structures. These, in turn, are validated using an a contrario approach that guarantees a reduced number of false detections. The use of a multiscale strategy permits the detection at different resolution levels, which makes the method robust to noise and blur.

Indoor scene understanding is an active research topic that has recently gained increasing attention in both computer graphics and computer vision community. This paper presents a novel automatic point cloud instance abstraction and segmentation of indoor scenes using learned pairwise planar patch relations. Planar patches have sufficient representative power to abstract the semantics in mostly man-made 3D indoor scenes. To exploit the planar patch relationships, a new indoor point clouds dataset is automatically derived from the public dataset ScanNet, which can be adopted for training a pairwise patch relation classifier. Given the pre-processed indoor scene point clouds, a set of planar patches are generated to provide robust and compact representation of indoor objects and parts. For each extracted planar patch, a feature descriptor in view of its geometric properties and appearance is calculated. Owing to the proposed feature descriptors, the relations of pairwise planar patch are measured and can be fed into the pre-trained patch relation classifier. To yield the instance segmentation of indoor scenes, a robust graph-based algorithm is adopted to group these planar patches which incorporate the learned patch relations. Experimental results demonstrate that our proposed approach can produce instance semantic segmentation of various cluttered indoor scenes robustly and efficiently.

One of the major shortcomings of existing image dehazing algorithms is in estimating scene transmittance, which has assumed many items in the existing algorithms. One key assumption has been pixel uniformity and smoothness. In this paper, we propose to solve the dehazing problem using a combination of single-pass CNN with graph cut algorithms. It considers the transmittance based on differential pixel-based variance, local and global patches and energy functions to improve the transmission map. The proposed algorithm was tested on different images and evaluated based on various evaluation metrics. Our results show more details when compared to four existing benchmark enhancement methods. The proposed method has one major drawback: the over-bright areas tend to lose some features in the final image.

Previous camera calibration methods often use a checkerboard to capture images and estimate the camera parameters from the correspondences between images and the checkerboard. The corner points in the checkerboard images are used as useful features for correspondence matching. Therefore, it is essential to precisely find the corner points in the checkerboard images. In many previous works, the corner points are extracted assuming that the checkerboard images are not distorted by its lens. Instead, image blurring and Gaussian noise on the images are usually considered, but other cases are not dealt with. However, the captured checkerboard images are often corrupted by lens distortions and compression artifacts, which leads to performance degradation of corner point extraction. Moreover, the corner points are extracted individually in the previous methods without considering their geometric relations. To better handle the corner point extraction problem under lens distortions, in our corner point extraction optimization, the distorted locations of the pixels on checkerboard images are corrected with the camera parameters, and the structural constraints for checkerboard image grids are then applied under the line-to-line mapping. Also, to robustly find the blurred edges between corner points due to JPEG compression, an edge surface model is newly proposed that models the transitions with over- and under-shoots around the blurred edges. Extensive experimental results show that our method significantly outperforms the state-of-the-art method with average 88.3% and 54.3% reduction in RMSE for corner point reprojection and camera parameter estimation, respectively under compression and lens distortions for synthetic and real data.

Aiming at the problems of texture distortion and fuzzy details in the existing image super-resolution reconstruction methods, a super-resolution reconstruction network based on multi-channel attention mechanism is proposed. The texture extraction module designs an extremely lightweight multi-channel attention module in the network structure. Combined with one-dimensional convolution, it realizes cross-channel information interaction, focusing on important feature information; The texture restoration module introduces dense remaining blocks to restore some high-frequency texture details, improve the model performance, and generate high-quality reconstructed images. The proposed network can not only effectively improve the visual effect of the image, but the results on the benchmark data set CUFED5 are similar to the classic super-resolution (SRCNN) reconstruction method based on convolutional neural network. The peak signal-to-noise ratio (PSNR) and structure are similar. Degrees (SSIM) increased by 1.76dB and 0.062 respectively. Experimental results show that the designed network can improve the accuracy of texture migration and can effectively improve the quality of the generated image.

Image feature detection is a key task in computer vision. Scale Invariant Feature Transform (SIFT) is a prevalent and well known algorithm for robust feature detection. However, it is computationally demanding and software implementations are not applicable for real-time performance. In this paper, a versatile and pipelined hardware implementation is proposed, that is capable of computing keypoints and rotation invariant descriptors on-chip. All computations are performed in single precision floating-point format which makes it possible to implement the original algorithm with little alteration. Various rotation resolutions and filter kernel sizes are supported for images of any resolution up to ultra-high definition. For full high definition images, 84 fps can be processed. Ultra high definition images can be processed at 21 fps.

We propose a single-shot high dynamic range (HDR) imaging algorithm with row-wise varying exposures in a single raw image based on a deep convolutional neural network (CNN). We first convert a raw Bayer input image into a radiance map by calibrating rows with different exposures, and then we design a new CNN model to restore missing information at the under- and over-exposed pixels and reconstruct color information from the raw radiance map. The proposed CNN model consists of three branch networks to obtain multiscale feature maps for an image. To effectively estimate the high-quality HDR images, we develop a robust loss function that considers the human visual system (HVS) model, color perception model, and multiscale contrast. Experimental results on both synthetic and captured real images demonstrate that the proposed algorithm can achieve synthesis results of significantly higher quality than conventional algorithms in terms of structure, color, and visual artifacts.

In this paper, we explore the problem of removing atmospheric turbulence to obtain better images of remote sidereal star systems. The imaging process is remodeled as a transmit–receive wireless communication paradigm in a novel approach for correcting space- and time-varying blur in stellar images. In particular, the effect of starlight passing through atmospheric turbulence is modeled as multipath fading communication channels. The problem is then transformed into channel equalization in space and time domains to produce a sharp stellar image. This approach involves first estimating the center of a blurred stellar image. Next, linear regression obtains an equivalent two-dimensional channel response through decomposition of the blurred image into the weighted sum of the diffraction-limited patterns in the spatial domain. Finally, an alignment algorithm is implemented for synthesis, and a final output is generated. Experiments were performed with real field-captured images; the results revealed that this approach could be used to effectively correct image blur and obtain diffraction-limited stellar images.

Traditional $\ell _{1}$ -regularized compressed sensing magnetic resonance imaging (CS-MRI) model tends to underestimate the fine textures and edges of the MR image, which play important roles in clinical diagnosis. In contrast, the convex nonconvex (CNC) strategy allows the use of nonconvex regularization while maintaining the convexity of the total objective function. Plug-and-play (PnP) algorithm is a powerful framework for sparse regularization problems, which plug any advanced denoiser into traditional proximal algorithms. In this paper, we propose a PnP-ADMM algorithm for CS-MRI reconstruction with CNC sparse regularization. We first obtain the proximal operator for CNC sparse regularization. Then we present PnP-ADMM algorithm by replacing the proximal operator of ADMM with properly pre-trained denoisers. Furthermore, we conduct comparison experiments using various denoisers under different sampling templates for different images. The experimental results verify the effectiveness of the proposed algorithm with both numerical criteria and visual effects.

Estimating the depth map from a single RGB image is important to understand the nature of the terrain in robot navigation and has attracted considerable attention in the past decade. The existing approaches can accurately estimate the depth from a single RGB image, considering a highly structured environment. The problem becomes more challenging when the terrain is highly dynamic. We propose a fine-tuned generative adversarial network to estimate the depth map effectively for a given single RGB image. The proposed network is composed of a fine-tuned generator and a global discriminator. The encoder part of the generator takes input RGB images and depth maps and generates their joint distribution in the latent space. Subsequently, the decoder part of the generator decodes the depth map from the joint distribution. The discriminator takes real and fake pairs in three different configurations and then guides the generator to estimate the depth map from the given RGB image accordingly. Finally, we conducted extensive experiments with a highly dynamic environment dataset for verifying the effectiveness and feasibility of the proposed approach. The proposed approach could decode the depth map from the joint distribution more effectively and accurately than the existing approaches.

Traditionally, multi-label feature extraction algorithm is used to extract local features of blurred images under complex illumination conditions. The computational complexity is extremely high and the extraction efficiency is low. Therefore, in this paper, the local feature intelligent extraction algorithm for blurred images under complex illumination conditions is proposed. The improved image segmentation algorithm based on local fuzzy C-means clustering is used to cluster and segment the blurred images. The image feature recognition algorithm based on wavelet transform and LBP log domain feature extraction is adopted. The blurred image is transformed from the spatial domain to the logarithmic domain to make two-stage discrete wavelet decomposition, and the high-frequency component is used to reconstruct the original image to perform high-pass filtering on the blurred image. By filtering the low-frequency illumination component to compensate the complex illumination, the block LBP is used to extract the local texture features of the blurred image after illumination compensation. The experimental results on Yale-B face database show that the proposed algorithm can effectively extract the local features of blurred images under complex illumination conditions, and the maximum robustness of the algorithm is 0.44, the maximum value of feature extraction error rate is 0.25, and the maximum value of feature extraction speed growth rate is 96%, with high robustness, accuracy and extraction efficiency.

This study designed a sub-pixel precision edge detecting algorithm to enhance contour smoothness. First, the coordinate value of RGB pixel is projected on the space line of R=G=B to obtain gray image. Then, pixel edges are located using a Canny detector. Next, the edge width is thinned to a single pixel using a morphological thinning operation. Finally, sub-pixel-level smooth contours are extracted by interpolation. In this sub-pixel level contour extraction process, a Single-Pixel-Multi-Point interpolation method was developed to enhance edge smoothness and obtain high precision in edge estimation. This method divides edges in a $3\times 3$ pixels block into nine arrangement modes. According to the arrangement of the eight neighborhoods of a centered edge pixel, different locations of interpolated sub-pixel points are calculated by interpolation with Bezier curves. For symmetrically arranged linear edge pixels, this method can be used to determine the exact contour. Experimental results showed that the proposed algorithm can improve the smoothness of image edge contour. As the curvature of the edge increases, the maximum systematic error will increase. For the edge pixel centered in the $3\times 3$ pixels block and two pixels located at the corner of one side, the max systematic error is 0.5 pixel. For two edge pixels aligned in a single row or column with one located at a corner, the max systematic error is 0.25 pixel.

As a nondestructive testing technology, $\gamma $ -photon imaging shows immense potential in the industrial field. However, the limitations of $\gamma $ -photon imaging theory and detection technology result in various problems, such as low image resolution and edge blur. The technology is particularly difficult to apply to industrial detection that requires high imaging speed and high resolution. Therefore, this study proposes a reconstruction algorithm for regions of interest (ROI) in $\gamma $ -photon images. The proposed algorithm is suitable for fast industrial detection and is based on the reconstruction algorithm for sinusoidal graph data, that is, the ordered subset expectation maximization (OSEM) image reconstruction algorithm. It is an improvement of the traditional point-and-line system matrix (SM) model. In the application of the proposed algorithm, the probability weight of a pixel is determined by the solid angle of the crystal bar at both ends of the line of response (LOR) to the pixel it passes through. In this work, the known contour parameters of industrial parts are used to describe the area of nuclide distribution as the ROI. Only the pixels through which the LOR passed in the ROI are counted, and the probability weights of these pixels are calculated to construct the SM. Gaussian filters are added in each iteration to suppress the clutter of scattered noise inside the image. The effectiveness of the algorithm was verified in two model experiments. A closed cavity detection experiment on industrial hydraulic parts was also conducted to compare the image reconstruction effects before and after the improvement. Results showed that the proposed algorithm can effectively improve image resolution and image edge contours. In the tee pipe model experiment and cavity detection experiment on hydraulic parts, the image reconstruction speed increased by more than 6 and 10 times, respectively. Hence, the proposed algorithm provided a feasible solution for quickly obtaining images with clear edges and high resolution under a large aperture detector ring.

The ever-evolving mesoscopic scale optical imaging systems facilitate the new discoveries in the neuroscience research. They excel at providing a complete view of the long projections of a single neuron across the whole brain. Although the preparation protocols and the optical imaging systems are rapidly evolving, there are still some noise and artifacts interfering downstream data processing and analysis due to the defects in the imaging system or during the sample preparation. A specific denoising procedure is usually developed for one optical imaging system as a post-processing measure. However, the development of the optical imaging systems usually follows in an incremental manner. It would be better to adapt the denoising model to the new optical imaging system than training a denoising model from scratch. In this paper, the proposed learning schema and practice learn a new denoising model for a new optical imaging system based on the existing denoise models and bootstrap a denoising model without the ground-truth denoising labels. We achieve this through a CycleGAN based model and the fact that the optical imaging systems usually produce images both from the signal area and the fixture area. The experiments show that our proposed procedure can provide a comparable denoising performance against other state-of-the-art denoising methods for several optical neuroimaging datasets.

For surface defect images that captured from a practical steel production line, different shape, size, location and texture of defect object may cause inter-class similarity and intra-class difference of defect images. Despite attractive results have been achieved in some surface methods for defect classification and segmentation, it is still far from meeting the needs of real-world applications due to lack of adaptiveness of these methods. Considering the surface defect image can be decomposed into defect foreground image and defect-free background image, the paper develops a novel joint classification and segmentation (JCS) approach to perform surface defects detection for steel sheet. It comprises of the classification method based on a class-specific and shared discriminative dictionary learning (CASDDL) and the segmentation method based on a double low-rank based matrix decomposition (DLMD), respectively. For the proposed CASDDL method, we learn a shared sub-dictionary as well as several class-specific sub-dictionaries to explicitly capture common information shared by all classes and class-specific information belonging to corresponding class. We adopt a mutual incoherence constrain for each sub-dictionary, a Fisher-like discriminative criterion and low-rank constrain on coding vector to improve the discriminative ability of learned dictionary. For the proposed DLMD method, we formulate the segmentation task as a double low-rank based matrix factorization problem, and the Laplacian and sparse regularization terms are introduced into the matrix decomposition framework. Experimental results demonstrate that our proposed JCS method achieve a comparable or better performance than the state-of- the-art methods in classifying and segmenting surface defects of steel sheet.

Photon counting imaging can be used to capture clearly photon-limited scenes. In photon counting imaging, information on incident photons is obtained as binary frames (bit-plane frames), which are transformed into a multi-bit image in the reconstruction process. In this process, it is necessary to apply a deblurring method to enable the capture of dynamic scenes without motion blur. In this article, a deblurring method for the high-quality bit-plane frame reconstruction of dynamic scenes is proposed. The proposed method involves the deblurring of units of object motion within a scene through the application of motion compensation to pixels sharing the same motions. This method achieves more efficient motion blur suppression than the application of simple deblurring to pixel block or spatial region units. It also applies a novel technique for accurate motion estimation from the bit-plane frame even in photon-limited situations through the statistical evaluation of the temporal variation of photon incidence. In addition to deblurring, our experimental results also revealed that the proposed method can be applied for denoising, which improves the peak signal-to-noise ratio by 1.2 dB. In summary, the proposed method for bit-plane reconstruction achieves high quality imaging even in photon-limited dynamic scenes.

ACMs have been demonstrated to be highly suitable as image segmentation models for computer vision tasks. Among other ACM, the local region-based models show better performance because they extract the local information regarding intensity in the neighborhood and embed it into the energy minimization function to guide the active contour to the boundary of the desired object. However, the online segmentation of noisy and inhomogeneous is still a challenging task for local region-based ACM models. To overcome this challenge, the paper proposes a novel region-based active contour model, named active contour model with local dilated convolution filter (ACLD). The ACLD integrates local image information in the form of a signed pressure force function. Then, a Gaussian kernel is applied using dilated convolution instead of discrete convolution for regularizing the level set formulation. Finally, instead of using a constant stopping condition, the ACLD automatically stops at the object boundaries. The proposed model shows improved image segmentation results visually combined with less computational time in the case of synthetic and natural images compared with the state-of-the-art models. Further, on the ISIC2017 dataset, the ACLD yields segmentation results with the highest accuracy.

This paper presents a new lossless image compression method based on the learning of pixel values and contexts through multilayer perceptrons (MLPs). The prediction errors and contexts obtained by MLPs are forwarded to adaptive arithmetic encoders, like the conventional lossless compression schemes. The MLP-based prediction has long been attempted for lossless compression, and recently convolutional neural networks (CNNs) are also adopted for the lossy/lossless coding. While the existing MLP-based lossless compression schemes focused only on accurate pixel prediction, we jointly predict the pixel values and contexts. We also adopt and design channel-wise progressive learning, residual learning, and duplex network in this MLP-based framework, which leads to improved coding gain compared to the conventional methods. Experiments show that the proposed method performs better than the conventional non-learning algorithms and also recent learning-based compression methods with practical computation time.

Various kinds of noise will be produced during the process of ultrasonic logging in high temperature and high-pressure environment under oil wells, which is blurring the logging image. This paper presents a novel end-to-end denoising model (ULNet) based on CNN and feature attention to address this problem and remove the noise from ultrasonic logging images. Our method mainly includes feature attention, feature enhancement based on residual model and reconstruction for ultrasonic logging image. Feature enhancement based on a residual model integrates global and local features to increase the expressive ability of the denoising model. Feature attention is used to distinguish the channel feature weights, and effective for blind denoising of actual images. Kernel dilation and skip-connection is used to reduce the computational cost during training. The Noise mapping results are used to reconstruct a clean image. Comprehensive quantitative and qualitative evaluations of results for selected study datasets collected at six oil wells in China show that this model is a feasible and effective means for denoising ultrasonic logging images. Overall, ULNet shows potential for practical ultrasonic logging images denoising.

Almost all image sensors measure only one color per pixel through the color filter array. Missing pixels are estimated using a demosaicing process. For this reason, a demosaiced image leaves a particular trace. When an image is manipulated or tampered, the demosaicing trace can be changed. This change can serve as a basic clue for detecting or localizing image tampering. Demosaicing pattern-based tampering localization algorithms require a re-interpolation process, and the prediction residue between the given image and the re-interpolated image is commonly used to localize tampered regions. However, the prediction residue is not always valid because the demosaicing interpolation kernel cannot be known, which deteriorates the localization performance. This paper presents an effective re-interpolation process using singular value decomposition for an unknown demosaicing method. First, the green channel of the given image is decomposed into four sub-images according to the Bayer pattern. For a small block of each sub-image, the singular value decomposition is performed. The prediction residue is obtained by reconstructing the image block after removing the largest singular value. The feature to localize the forged regions is extracted by the logarithm ratio of the prediction residue variance. The proposed method does not require any statistical model for the extracted feature, because the prediction residue is more accurate than that of conventional methods. We perform intensive experiments for three test datasets and compare the proposed method with state-of-the-art tampering localization methods, the results of which indicate that the proposed scheme outperforms existing approaches.

To improve the fusion quality of infrared and visible images and highlight target and scene details, in this paper, a novel infrared and visible image fusion algorithm is proposed. First, a method for combining dynamic range compression and contrast restoration based on a guided filter is adopted to enhance the contrast of visible source images. Second, guided filter-based image multiscale decomposition is used to decompose images into base layers and detail layers. For base layer fusion, a fusion strategy based on the detail and energy measurements of the source image is proposed to determine the pixel value of the fused image base layer such that the energy loss of the fusion can be reduced and the texture detail features are highlighted to obtain more source image details. Finally, recursive separation and weighted histogram equalization methods are applied to optimize the fused image. Experimental results show that the fusion algorithm and fusion strategy proposed in this paper can effectively improve fusion image clarity, while more detailed target and scene information can still be retained.

The whiskbroom scanning imaging mode effectively expands the imaging field of the space optical camera, and realizes high-resolution and wide-width spatial imaging. However, the dynamic disturbance in the short-term and large-scale imaging process will cause uneven degradation of the image. Therefore, a geometric model of the camera instantaneous imaging at a moment is constructed. Then, the movement trend of the camera and the deflection of the light are deduced separately for establishing the analysis model of the kinematic disturbance and the morphological disturbance. Finally, the dynamic disturbance result of the imaging process is obtained by fusing both kinematic disturbance and morphological disturbance, and the simulated image of the dynamic disturbance is evaluated by combining the geometric parameters and the image quality evaluation index. The results show that the dynamic disturbance of the imaging will be enhanced with the increase of the whiskbroom angle and angular velocity, resulting in the decrease of the peak signal-to-noise ratio and the structural similarity of the image. When the angular velocity on the 500km orbit increases from 1°/ $s$ to 10°/ $s$ , the peak signal-to-noise ratio and structural similarity of the simulated image, affected by the combining influence of kinematic disturbance and morphological disturbance, decrease by 2.1454 dB and 0.1131 respectively at 0°, but decrease by 1.5680 dB and 0.0405 at 45°. The analysis results of this paper can provide an effective reference for further research on the degradation process of whiskbroom ultra-wide image and improve the quality of the image.

By combining the basics of self-similarity, scaling correlation, and statistical components, Benoit Mandelbrot formulated the idea of a natural fractal entity, an entity described by those fundamentals. As a result of these principles, fractal image codings are being used in many substantial applications already, such as image compression, image signature, image watermarking, image attribute extraction, and even image texture segmentation. Thus, while fractal image coding is relatively new in the field of image encoding, it has gained broad acceptance at a rapid pace. In light of its beneficial qualities, such as quick decomposition, high compression ratio, and the independence of resolution at any size make these applications conceivable. However, compared to its advantages, fractal image coding is extremely time-complex and so remarkably expensive, which hinders its prevalence. A wide hunting domain blocks for the relevant range blocks caused this difficulty. We proposed several improvements to the Jacquin design in this paper. We first used max-pooling as an alternative for the medium bonding of spatial contractions to validate the value of the edge textures of the block. Secondly, we construct the odd-size pixel block alternative to an even-size pixel block for validation of the symmetric central pixel (CP). Finally, before the search started, we proposed a shortening of block space, using the central pixel of the block to convert each eight-bit pixel to a two-bit pixel. As a consequence, the symmetrical CP of odd pixels block, reduction of block space, and edge pixel selection accomplished faster coding and competitive image quality than existing known exhaustive search algorithms.

Generally, most existing super-resolution (SR) methods do not consider noise, which treats SR reconstruction and denoising as two separate problems and performs separately. However, noise is inevitably introduced in the imaging process. Based on analysis of the degraded model, in this paper, the problems of interpolation and denoising are modeled to estimate the noiseless and missing images under the same framework. By applying local fractal dimension (LFD) into image local feature analysis, a noisy single-image SR method is proposed. For each noisy image, we first construct a rational fractal interpolation model containing scaling factors, which can effectively maintain the inherent properties of the data. Furthermore, the original image structure can be well preserved by applying the interpolation model. Considering the local characteristics of the image, scaling factors are calculated on the basis of the LFDs. Then, through further local feature analysis of the interpolated image, a denoising method based on LFD is proposed for recovering a noiseless image. Finally, a high-quality high-resolution image is obtained. Experimental results demonstrate that our method outperforms the state-of-the-art methods both quantitatively and qualitatively.

Denoising plays an important role in the Magnetic Resonance Imaging (MRI) applications for medical diagnosis. MRI images usually contain undesired noises which would negatively affect the exactitude of pathological diagnosis. Recently, many models for MRI denoising have been developed from deep learning networks. In this paper, we propose a novel MRI image denoising method using the conditional Generative Adversarial Networks (GANs). Specifically, a Convolutional Neural Network (CNN) is utilized as the discriminator in the process to distinguish whether the image pair obtained from the conditional GAN is a real pair which consists of a noisy image and a noise-free image or a fake pair which, on the other hand, contains a noisy image and a denoised image. In our design, the convolutional encoder-decoder networks-based generator is used to remove the noise in the noisy MRI images as much as possible. The whole architecture is trained by adversarial learning. Experiments using both synthetic and real clinical MRI datasets are conducted. When tested on the synthetic T1w images with 10% noise level, our method performed better in terms of reaching a high structural similarity index (SSIM) at 0.9489 while that of the next best method was only 0.7485. Moreover, when the image noise level was increased from 1% to 10%, our method was more stable that the SSIM only dropped about 3.2% while that of the next best method dropped about 23.7%. Simulation results demonstrate that the proposed method is more robust and outperforms the conventional methods in both the denoising level and preservation of the anatomical structures and defined contrast.

In this paper, a noise-resistant image encryption scheme is proposed. We have used a cubic-logistic map, Discrete Wavelet Transform (DWT), and bit-plane extraction method to encrypt the medical images at the bit-level rather than pixel-level. The proposed work is divided into three sections; In the first and the last section, the image is encrypted in the spatial domain. While the middle section of the proposed algorithm is devoted to the frequency domain encryption in which DWT is incorporated. As the frequency domain encryption section is a sandwich between the two spatial domain encryption sections, we called it a ”sandwich encryption.” The proposed algorithm is lossless because it can decrypt the exact pixel values of an image. Along with this, we have also gauge the proposed scheme’s performance using statistical analysis such as entropy, correlation, and contrast. The entropy values of the cipher images generated from the proposed encryption scheme are more remarkable than 7.99, while correlation values are very close to zero. Furthermore, the number of pixel change rate (NPCR) and unified average change intensity (UACI) for the proposed encryption scheme is higher than 99.4% and 33, respectively. We have also tested the proposed algorithm by performing attacks such as cropping and noise attacks on enciphered images, and we found that the proposed algorithm can decrypt the plaintext image with little loss of information, but the content of the original image is visible.

Ghost imaging reconstructs images using a single-element photodetector; it performs imaging by illuminating an object with binary modulation patterns. This technique has various advantages, including a wide wavelength, noise robustness, and high measurement sensitivity. However, one challenge is the low image quality in the undersampling. The examination of modulation patterns is intended to solve this issue. In ghost imaging, randomly generated or basis patterns have been studied as modulation patterns; however, in undersampling, random patterns exhibit noise robustness but low image quality, whereas basis patterns exhibit high image quality but are sensitive to noise and low resolution. Thus, ghost imaging requires patterns that simultaneously achieve high image quality, resolution, and noise robustness. This study proposes a method of pattern optimization using gradient descent and a binarization method to further improve image quality. Numerical simulation and experimental results show that the proposed approach offers high image quality, high resolution, and robustness to noise.

As a nondestructive testing technology, $\gamma $ -photon imaging shows immense potential in the industrial field. However, the limitations of $\gamma $ -photon imaging theory and detection technology result in various problems, such as low image resolution and edge blur. The technology is particularly difficult to apply to industrial detection that requires high imaging speed and high resolution. Therefore, this study proposes a reconstruction algorithm for regions of interest (ROI) in $\gamma $ -photon images. The proposed algorithm is suitable for fast industrial detection and is based on the reconstruction algorithm for sinusoidal graph data, that is, the ordered subset expectation maximization (OSEM) image reconstruction algorithm. It is an improvement of the traditional point-and-line system matrix (SM) model. In the application of the proposed algorithm, the probability weight of a pixel is determined by the solid angle of the crystal bar at both ends of the line of response (LOR) to the pixel it passes through. In this work, the known contour parameters of industrial parts are used to describe the area of nuclide distribution as the ROI. Only the pixels through which the LOR passed in the ROI are counted, and the probability weights of these pixels are calculated to construct the SM. Gaussian filters are added in each iteration to suppress the clutter of scattered noise inside the image. The effectiveness of the algorithm was verified in two model experiments. A closed cavity detection experiment on industrial hydraulic parts was also conducted to compare the image reconstruction effects before and after the improvement. Results showed that the proposed algorithm can effectively improve image resolution and image edge contours. In the tee pipe model experiment and cavity detection experiment on hydraulic parts, the image reconstruction speed increased by more than 6 and 10 times, respectively. Hence, the proposed algorithm provided a feasible solution for quickly obtaining images with clear edges and high resolution under a large aperture detector ring.

International audience; Spectral photon-counting computed tomography (sCT) appears as a promising imaging technique for clinical applications thanks to its ability to offer low dose and possibility of quantitatively analyzing the composition of materials in a pixel. However, due to the dispatching of photons into different energy bins, the quality of sCT image at each energy bin is considerably degraded. We propose a reconstruction method for sCT images by combining multi-energy information. The method is based on clustering pixels containing similar material compositions, performing linear fitting within each class for all the energy images two-by-two, projecting the pixel values of the images at other energy bins to the pixel of the image at the current energy bin, and combining the original pixel value and projected pixel values. The results on both simulation and real data demonstrated the effectiveness of the proposed method, in terms of both image reconstruction quality and material decomposition accuracy.

In order to completely separate objects with large sections of occluded boundaries in an image, we devise a new variational level set model for image segmentation combining the Chan-Vese model with elastica and landmark constraints. For computational efficiency, we design its Augmented Lagrangian Method (ALM) or Alternating Direction Method of Multiplier (ADMM) method by introducing some auxiliary variables, Lagrange multipliers, and penalty parameters. In each loop of alternating iterative optimization, the sub-problems of minimization can be easily solved via the Gauss-Seidel iterative method and generalized soft thresholding formulas with projection, respectively. Numerical experiments show that the proposed model can not only recover larger broken boundaries but can also improve segmentation efficiency, as well as decrease the dependence of segmentation on parameter tuning and initialization., 17pages,9figure

Impressive progress has been recently witnessed on deep unsupervised clustering and feature disentanglement. In this paper, we propose a novel method on top of one recent architecture with a novel explanation of Gaussian mixture model (GMM) membership, accompanied by a GMM loss to enhance the clustering. The GMM loss is optimized with the explicitly computed parameters under our coupled GMM inspired framework. Specifically, our model takes the advantage of implicitly learning a GMM in latent space by neural networks (GMM prior as the first GMM), and explicitly clustering via the other GMM framework (GMM estimator as the second GMM). We further introduce a Dirichlet conjugate loss as a regularization term to prevent the GMM estimator from degenerating to few Gaussians. Eventually, we further propose an application of apparel generation based on the proposed method which requires only three selection steps. Extensive experiments on publicly available datasets demonstrate the effectiveness of our method, in terms of clustering and disentanglement performance.

The key to the depth estimation from a single image lies in inferring the distance of various objects without copying texture while maintaining clear object boundaries. In this paper, we propose depth estimation from a single image using edge extraction network and dark channel prior (DCP). We build an edge extraction network based on generative adversarial networks (GANs) to select valid depth edges from a number of edges in an image. We use DCP to generate a transmission map that is able to represent distance from the camera. Transmission map is generated by conducting minimum value filtering on DCP. First, we concatenate the transmission map with the original RGB image to form a tensor, i.e. RGB + T. Second, we generate an initial depth image from the tensor through the generator by inferring depth from stacked residual blocks. Third, we compare the edge map of the initial depth image with that of the input RGB image to select valid depth edges. Both edge maps are generated by the edge extraction network. Finally, we distinguish real and fake on the generated depth image using a discriminator, and enhances the performance of the generator. Various experiments on NYU, Make3D and MPI Sintel datasets demonstrate that the proposed network generates clear edges in depth images as well as outperforms state-of-the-art methods in terms of visual quality and quantitative measurements.

Cryptography is one of the most important security mechanisms for the transmission of digital media on the Internet. Most proposed cryptographic image methods based on chaotic maps are dependent upon image sizes and most of them worked on square images. In this paper for tackling this problem, a scale-invariant color image encryption method in three-dimensional space is presented. At first, the two-dimensional color image is converted into three-dimensional space, in this case, the red, green, and blue color spectrums are divided into a set of gray-level square sub-images. Then, to have confusion and diffusion properties, the 3D substitution and 3D permutation operation are performed on the sub-images. In substitution operations, the pixel values of the sub-images are changed with the help of XOR and circular shift operators with appropriate keys. In permutation operation, the position of the pixels is changed using modular three-dimensional chaos mappings. To have scale-invariant three-dimensional permutation, the sub-images are divided into one or more windows with equal size, and then 3D modular chaotic map operations are performed on each window with separate keys. Depending on the number of sub-images, there may be two last windows that have overlapping. To increase the speed of color image encryption, the steps of a 3D modular chaotic map on the windows can be implemented in parallel. The proposed approach relative to the similar color image encryption algorithms increases the key space and improves standard parameters, such as entropy, sensitivity, adjacent pixel correlations, and histogram uniformity.

Registration of multi-modal images is one of the challenging problems in image processing nowadays. In this paper, two novel non-rigid registration models are proposed for multi-modality images. In model 1, mutual information of the template and reference images is used as data fitting term with Gaussian curvature regularization. This approach may not give satisfactory results in noisy images or images having bias field. To overcome this drawback, model 2 is proposed which is based on normalized gradient of both template and reference images as a data fitting term instead of mutual information. To get best transformations, both the models are minimized by using Augmented Lagrangian Method. The proposed models can register multi-modality images without effecting edges and other important fine details and are also tested on various medical images like (T1-T2 MRI, PD weighted-T2 MRI) noisy and synthetic images. The proposed models are also tested on a well known free available Brainweb dataset, where they produced satisfactory results. From experimental results, it can be observed that normalized gradient field based model gives better results than mutual information based model. Comparison is done qualitatively and quantitatively through Jaccard Similarity Coefficient.

To address the problems of edge blur and weak detail resolution when fusing infrared and visible images with traditional methods, a novel image fusion approach based on fast super-resolution convolutional neural network and anisotropic diffusion is proposed. The proposed method first employs super-resolution reconstruction based on fast convolutional neural networks to get high-resolution infrared images. Second, the visible and high-resolution infrared images are anisotropically diffused to obtain the corresponding base layers and detail layers. Next, the fusion of the detail layer is performed through Karhunen-Loeve transform, and an adaptive saliency module is constructed to fuse the base layer. Finally, the fusion result is obtained by applying linear reconstruction on the fused detail layer and base layer. Compared with other algorithms, the experimental results show that the proposed algorithm not only has better objective evaluation indexes but also can better retain the details of the original image.

With the development of computational photography, single-lens camera combined with corresponding image deblurring algorithm is gradually becoming a new research direction, replacing complex modern optical imaging system such as single lens reflex (SLR) camera. For single-lens camera, the Point Spread Function (PSF) estimation accuracy will directly affect the image restoration effect. In this paper, we designed the simple-lens cameras with one, two and three lenses, respectively, and propose a robust and accurate PSF estimation method of simple-lens camera. The key point of estimation is to obtain the blur image and clear image pairs, which are necessary for non-blind deconvolution PSF estimation. Considering the structure characteristic of simple-lens camera, we take picture of original clear image displayed on the computer screen to get the image pairs through corner detection and color correction is made to remove color distortion. In addition, a few studies have shown that the PSF of the simple lens is close to the spatially deformed wedge, so we use a more reasonable Normal Sinh-Arcsinh (NSAS) model to fit the blur kernel and get its parameters by Powell algorithm. The experiment results have shown that the space-variant PSF estimated by the proposed method achieves better performance than the compared methods both qualitatively and quantitatively.

In recent years, convolutional neural networks have achieved considerable success in different computer vision tasks, including image denoising. In this work, we present a residual dense neural network (RDUNet) for image denoising based on the densely connected hierarchical network. The encoding and decoding layers of the RDUNet consist of densely connected convolutional layers to reuse the feature maps and local residual learning to avoid the vanishing gradient problem and speed up the learning process. Moreover, global residual learning is adopted such that, instead of directly predicting the denoised image, the model predicts the residual noise of the corrupted image. The algorithm was trained for the case of additive white Gaussian noise and using a wide range of noise levels. Hence, one advantage of the proposal is that the denoising process does not require prior knowledge about the noise level. In order to evaluate the model, we conducted several experiments with natural image databases available online, achieving competitive results compared with state-of-the-art networks for image denoising. For comparison purpose, we use additive Gaussian noise with levels 10, 30, 50. In the case of grayscale images, we achieved PSNR of 34.39, 29.11, 26.99, and SSIM of 0.9297, 0.8193, 0.7491. For color images we obtained PSNR of 36.68, 31.43, 29.12, and SSIM of 0.9600, 0.8961, 0.8465.

A simple fingerprint identification scheme compares an input fingerprint with all the fingerprints in the database to find any matching fingerprint. That is, the simple matching method considers all fingerprints in the database as candidates for a given input fingerprint. However, this simple matching method requires a lot of processing time. To reduce the processing time, we can use fingerprint indexing to reduce the size of a candidate set for an input fingerprint. The candidate set is the set of fingerprints most similar to the input fingerprint. Usually, the size of the candidate set is much smaller than the size of the whole fingerprint database. It enables efficient identification by comparing the input fingerprint with only the fingerprints in the candidate set instead of the entire database. In this paper, we analyze the index distribution of the Kavati et al. ’s indexing method and propose a new fingerprint index vector which tries to make the index distribution more similar to the uniform distribution. Our new index vector consists of elements that are not highly correlated, which is measured by the Pearson correlation coefficient. Because our indexing method makes the index values widely spread over the index space, it reduces the number of candidates for a given fingerprint in fingerprint identification. Our indexing method shows a higher match rate with a smaller candidate set than the existing triplet-based indexing methods. Especially, our indexing method is up to 6.4 times more accurate than the Kavati et al. ’s indexing method. Our result shows that the index distribution significantly affects performance of indexing methods.

Shape from Focus (SFF) has been studied extensively in computer vision for 3D shape and depth recovery. The first stage in SFF methods is to compute the focus value of every pixel by converting the colored images into gray scale and then apply the focus measure operator. Converting colored values in the images into gray scale values may lead to imprecise mapping of pixels with different colored values onto the same gray scale value, this affects the overall accuracy of the system. In a colored image, the focused pixels maintain a considerable color difference from their neighboring pixels as compared to the defocused ones, which are blended into their neighborhood. This article presents an alternative method to measure the degree of focus by directly processing colored images. The color differences of the neighbor pixels with respect to the central pixel are obtained and summed together, this is followed by calculating their spread. The sum and the spread are combined to measure the degree of focus of the pixel in consideration. The proposed focus measure is then used for shape recovery of various simulated and real objects and is compared with previous techniques. The comparison results show the proposed method has the highest correlation and smallest RMSE values confirming the effectiveness of using color images for shape recovery.

The ever-increasing resolution puts tremendous pressure to the onboard hyperspectral imaging system. Compressed sensing technology is one of the important ways to solve this problem. Distributed compressed sensing was proposed to exploit both intra- and inter-correlation structures of hyperspectral images. However, the implementation method of distributed compressed sampling has not been reported, and the joint sparsity reconstruction algorithm cannot achieve excellent image reconstruction performance. In this paper, a distributed compressed sampling strategy inspired by distributed compressed video sensing and optical implementation model are proposed to collect compressed hyperspectral data. In the image reconstruction process, we discard the direct application of the joint sparsity constraint on the data itself. Instead, we explore the estimation method of abundance and endmember with the help of the existing spectral library. Then, the images are recovered by applying the linear mixing model of hyperspectral. The comparison experiments of various schemes show that the proposed compressed sensing scheme has an obvious advantage in reconstruction performance under the low sampling rate. The proposed compressed sensing scheme has great potential in a high-compression ratio onboard hyperspectral imaging system.

Magnetic resonance imaging (MRI) is a medical technology that uses powerful magnets, radio waves, and a computer to produce images of the body’s internal organs. The patient should be quiet and motionless during the scanning period, as unavoidable movements, such as breathing and heart rate, cause motion artifacts in the image, which cause contrast instability and low-resolution MRI images. Imaging in the clinical setting is performed at low resolution because the scanning time for high-resolution MR imaging would be very long and cumbersome, and imaging is also very expensive. Learning-based image superresolution methods can reconstruct MR images with optimal quality and resolution. However, these methods have problems such as the inability to find the intrinsic relationship between low- and high-resolution image patches in the training dictionary, specification of the proper amount of error in the training and testing stage due to variability in MR image contrast, inability to reconstruct objects by smoothed edges, and use of the backpropagation method in updating their weight vectors. In this paper, we propose a new superresolution method according to competitive learning-based approaches to overcome the problems of previous superresolution methods, which do not have the problems and complexities of those methods. The proposed method includes self-organizing maps as a preprocessor, the nearest neighbor algorithm as a classifier, and a high-frequency filter as a high-frequency image detail extractor. We constructed a single external dictionary from a combination of low-resolution and high-resolution feature patches and trained our SOM network. Next, we reconstruct the high-resolution image by converting the low-resolution input image into feature patch vectors, and for each vector, we find all corresponding neurons in the network and retrieve all their training feature vectors. Next, we train the nearest neighbor algorithm with the recovered vectors plus the input vector and find the best similarity vector to the input vector. After finding all the best similarity vectors to the input vectors, we reconstruct our high superresolution image. The proposed image superresolution method in practical experiments was trained, tested, and evaluated by the Div2k dataset and compared with other traditional and state-of-the-art image enhancement methods on various datasets, such as SET5, SET14, BSDS100, and URBAN100, and presented better results with higher accuracy and quality than traditional and state-of-the-art methods, both visually compared to each other by human and computational benchmarks, such as the peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM), to compare image superresolution algorithms. This method is best for reconstructing high-resolution magnetic resonance images that require high-frequency details and sharp edges with a smooth slope of the imaging objects in their structures. The execution time of the proposed algorithm is slower than that of the other algorithms, so we use GPU hardware and parallel programming techniques to increase the algorithm speed.

In this paper, a new model combining four-directional total variation with overlapping group sparsity is proposed, which not only suppresses the staircase effects introduced by traditional total variation, but also fully utilizes the gradient neighborhood information on each pixel of the image. In order to decrease the computation time of image denoising, the alternating direction method of multipliers (ADMM) is adopted to divide the complex optimization problem into separate subproblems that are easy to solve. At the same time, two-dimensional Fast Fourier Transform (FFT) and majorization-minimization (MM) are used to solve the subproblems alternatively. Then, the proposed new model is compared with other state-of-the-art models. Experiments show that the new model is robust in denoising. The new model not only excavates the gradient information of the four directions on the image to remove the noise more effectively, but also better in preserving image features, further reducing staircase artifacts.