Search

The performance of vision systems can be affected when used in severe weather conditions such as heavy rain or snow. Rain streak removal is an ill posed problem as they can vary in shape, size, and density across the image. In this paper, a single-image deraining network named Factorized Multi-scale Multi-resolution Residual Network (FMMRNet), which follows a U-Net backbone, is proposed. As rain streaks affect non-local regions of the image, larger receptive fields are beneficial to capture these non-local dependencies. We propose the use of multi-scale grouped convolutions to integrate information from global and local scales. In the proposed FMMRNet, multi-scale convolutions are factorized so that they can have large effective kernel sizes while reducing the computational complexity. During training, intermediate multi-resolution outputs are produced for loss computation which improves gradient flow in the deeper layers of the network and promotes better learning. A channel-wise attention mechanism is included to recalibrate feature maps before fusion instead of direct concatenation at each stage of the decoder. A higher-level reconstruction loss called perceptual loss is included for effective training to improve the visual quality of the derained images. The performance of the proposed FMMRNet is quantitatively and qualitatively compared on public benchmark datasets, and it outperforms the state-of-the-art deraining methods. Furthermore, to show the practical applicability of the proposed network, we demonstrate that when FMMRNet is used as a preprocessing step for object-detection methods such as Faster-RCNN and YOLO, it improves their performance on images degraded by rain streaks by 12.6% and 75.2% respectively.

Recently, deep convolutional neural networks have achieved great success for single-image deraining. However, affected by the intrinsic overlapping between rain streaks and background texture patterns, a majority of these methods tend to almost remove texture details in rain-free regions and lead to over-smoothing effects in the recovered background. To generate reasonable rain streak layers and improve the reconstruction quality of the background, we propose a distributed feedback network (DFN) in recurrent structure . A novel feedback block is designed to implement the feedback mechanism. In each feedback block, the hidden state with high-level information (output) will flow into the next iteration to correct the low-level representations (input). By stacking multiple feedback blocks, the proposed network where the hidden states are distributed can extract powerful high-level representations for rain streak layers. Curriculum learning is employed to connect the loss of each iteration and ensure that hidden states contain the notion of output. In addition, a self-ensemble strategy for rain removal task, which can retain the approximate vertical character of rain streaks, is explored to maximize the potential performance of the deraining model. Extensive experimental results demonstrated the superiority of the proposed method in comparison with other deraining methods.

Introduction Deep learning (DL) is used to classify, detect, and quantify gold nanoparticles (AuNPs) in a human-sized phantom with a clinical MDCT scanner. Methods AuNPs were imaged at concentrations between 0.0274 and 200 mgAu/mL in a 33 cm phantom. 1 mm-thick CT image slices were acquired at 120 kVp with a CTDIvol of 23.6 mGy. A convolutional neural network (CNN) was trained on 544 images to classify 17 different tissue types and AuNP concentrations. A second set of 544 images was then used for testing. Results AuNPs were classified with 95% accuracy at 0.1095 mgAu/mL and 97% accuracy at 0.2189 mgAu/mL. Both these concentrations are lower than what humans can visually perceive (0.3–1.4 mgAu/mL). AuNP concentrations were also classified with 95% accuracy at 150 and 200 mgAu/mL. These high concentrations result in CT numbers that are at or above the 12-bit limit for CT’s dynamic range where extended Hounsfield scales are otherwise required for measuring differences in contrast. Conclusions We have shown that DL can be used to detect AuNPs at concentrations lower than what humans can visually perceive and can also quantify very high AuNP concentrations that exceed the typical 12-bit dynamic range of clinical MDCT scanners. This second finding is possible due to inhomogeneous AuNP distributions and characteristic streak artifacts. It may even be possible to extend this approach beyond AuNP imaging in CT for quantifying high density objects without extended Hounsfield scales.

This work presents a gated non-local deep residual learning framework for image deraining. It can avoid the over-deraining or under-deraining caused by the global residual learning in existing deraining networks, since the learned soft gate in our method adaptively adjusts the amount of global residual to be passed for generating the final derained result. To generate feature maps for global residual prediction, we develop a non-local guided attention module (NLAM), which first obtains non-local features by exploiting spatial inter-dependencies among all the feature positions of local features produced by convolutional neural network (CNN), and then leverages the attention mechanism to merge the local and non-local features based on their complementary relation. Moreover, we develop a channel-wise gated prediction module to learn a soft gate on the global residual by explicitly modelling channel inter-dependencies of the feature maps obtained from NLAM. Experiments on four deraining benchmark datasets and real-world rainy images show that our network has a quantitative and qualitative improvement over state-of-the-arts.

Artifacts arising when differential phase images are integrated is a common problem to several X-ray phase-based experimental techniques. The combination of noise and insufficient sampling of the high-frequency differential phase signal leads to the formation of streak artifacts in the projections, translating into poor image quality in the tomography slices. In this work, we apply a non-iterative integration algorithm proven to reduce streak artifacts in planar (2D) images to a differential phase tomography scan. We report on how the reduction of streak artifacts in the projections improves the quality of the tomography slices, especially in the directions different from the reconstruction plane. Importantly, the method is compatible with large tomography datasets in terms of computation time.

The quality of photos is highly susceptible to severe weather such as heavy rain; it can also degrade the performance of various visual tasks like object detection. Rain removal is a challenging problem because rain streaks have different appearances even in one image. Regions where rain accumulates appear foggy or misty, while rain streaks can be clearly seen in areas where rain is less heavy. We propose removing various rain effects in pictures using a hybrid multiscale loss guided multiple feature fusion de-raining network (MSGMFFNet). Specially, to deal with rain streaks, our method generates a rain streak attention map, while preprocessing uses gamma correction and contrast enhancement to enhanced images to address the problem of rain accumulation. Using these tools, the model can restore a result with abundant details. Furthermore, a hybrid multiscale loss combining L1 loss and edge loss is used to guide the training process to pay attention to edge and content information. Comprehensive experiments conducted on both synthetic and real-world datasets demonstrate the effectiveness of our method.

Single image rain streak removal is a significant and challenging task, which is widely applied in many artificial intelligence domains as preprocessing process. Most of existing rain streak removal works focus on designing various deraining unit (e.g., multi-stream dilation convolution) without considering the correlation between different convolution layers, which may lead to large model size. In this paper, we propose a simple and effective deep network architecture for single image rain streak removal based on deep Convolutional Neural Network (CNN). Benefit from the adjacent layers with different dilation factors have similar feature structures, we design a powerful rain streak representation network based on the Layer Similarity Prior Block (LSPB). To better cater to the property of layer similarity prior, the multi-dense-short-connection is developed and it regards every LSPB as a convolution layer, this connection style makes our overall framework to be a layer similarity prior network. To the best of our knowledge, this is the first paper to investigate the effectiveness of exploiting the layer similarity prior and the multi-dense-short-connection. Quantitative and qualitative experimental results demonstrate that the proposed method outperforms other state-of-the-art methods with the least parameters.

In rainy conditions, imaging devices often capture degraded and blurry images. Most existing works on this problem focus on rain streak removal, but these approaches cannot handle the various types of rain found in images. In this paper, we propose a robust rain removal method for use with single images using an attentive composite residual network. We put forth a single-to-dual encoder-decoder structure, which consists of an attentive net that identifies regions containing rain components during encoding, followed by a dual-channel architecture which recovers the background and detail components of the identified regions during decoding. For the detail subnet, we designed a novel building block, namely a composite residual block (CRB), by constructing multiple residual connections among Res2Net modules. Additionally, we designed another attentive-CRB for the attentive net that uses a squeeze-and-excitation (SE)-Res2Net module, to build a channel-wise attention mechanism. We show that such a deep network can be trained end-to-end from rainy images and that it outperforms the previous state-of-the-art methods on datasets containing different types of rainy images. Experimental results also demonstrate the proposed model's superiority over the competitor's on real rain-affected images, recovering visually clean images and retaining good detail.

The major task of traditional motion deblurring methods is to estimate the blur kernel and restore the latent image. In low-light conditions, the pointolite is likely to produce saturated light streaks in captured blurred images. The light streaks are usually double-edged swords—outliers to the deconvolution, but a cue to kernel estimation. In this paper, we propose a novel blind motion deblurring method for blurred images including light streaks. The main idea is to model the non-linear blur caused by outliers as the Huber's M-estimation in blind deconvolution and take the shape of the light streak as a cue to estimate the blur kernel. Specifically, the optimal light streak patch is selected automatically according to the characteristics of light streaks and the blur kernel. This simple yet effective selection strategy solves the problems of false detection of candidate light streaks and optimal light streak in existing methods. Then, the optimal light streak patch is parameterized as a prior and is combined with other regularizers to estimate the blur kernel. Compared with the state-of-the-art kernel estimation methods, the proposed algorithm reduces the influence of outliers on deconvolution and utilizes more information. Thus, the restored image is more accurate. Experimental results on both synthetic and real images demonstrate the high accuracy of our algorithm.

This paper proposes a reverse procedure detection method for space target streaks based on motion parameter estimation. According to the phase shift characteristics of the spectrum, the interframe phase difference is used to obtain the displacement vectors of the target streak and image background, and then the morphological parameters of the target streak are estimated. In addition, a streak window is designed on the basis of the parameters, utilizing the local grayscale correlation in two image frames to carry out target searching and positioning with the interval of the target displacement vector. Images from different scenes are used to evaluate the adaptability of our method, and the experimental results suggest that the expected targets can be detected successfully with subpixel precision in all telescope operating modes. The proposed frequency-domain reverse procedure detection method avoids the problem of fake target confirmation caused by blind detection in traditional methods and overcomes the interference produced when the target streak passes through a bright star. Moreover, first target tracking and then positioning, which cannot be performed with time-domain methods, can be achieved. Our proposed method is attractive for multiple applications, such as space target cataloging, early warning and orbit prediction.

Low-dose computed tomography (LDCT) image often contains mottle noise and streak artefacts, which seriously interfere with clinical diagnosis. In this study, the separation-based (SEPB) method is proposed for mottle noise and streak artefacts suppression and structure preservation. In it, the LDCT image is decomposed into the structural image with residual mottle noise and the streak artefacts image with residual structural details by the image decomposition structural-preserving image smoothing method. The structural image is filtered by the K-singular value decomposition algorithm to remove the residual mottle noise, and the structural details in the streak artefacts image are extracted by the morphological component analysis theory. The extracted structural details are added to the filtered structural image to get the LDCT result image. Meanwhile, in the process of extracting the structural details, the streak artefacts dictionary learned from the streak artefacts image is corrected by the local intuitional fuzzy entropy to remove its structural atoms. The experiments are conducted on the modified Shepp–Logan phantom, the pelvis phantom and the clinical abdominal data to evaluate the proposed SEPB method. Compared to several comparative denoising methods, the experimental results show that the SEPB method has better performance in subjective visual effect and objective indicators.

High-quality image de-raining is a challenging task that has been given considerable importance in recent times. To begin with, this problem is modeled as an image decomposition task where a rainy image is decomposed into the rain-free background and the associated rain streak map. Most of the existing methods have been successful in removing the rain-streaks but fails to restore the image quality, which is degraded due to noise removal. This paper proposes a novel architecture called High-Resolution Image De-Raining using Conditional Generative Adversarial Networks (HRID-GAN) to generate a de-rained image with minimal artifacts and better visual quality. Extensive experiments on publicly available synthetic as well as real-world datasets show a substantial improvement over the state-of-the-art methods SPANet (Wang et al. 2019) by ∼ 2.43% in PSNR and, DID-MDN (Zhang and Patel 2018) by ∼ 2.43%, ∼ 10.12% and ID-CGAN (Zhang et al. 2017) by ∼ 11.80%, ∼ 34.70% in SSIM and PSNR respectively.

We detect the trajectory of particles using the feature matching method to improve the resolution of particle streak velocimetry (PSV), which is used to measure the velocity of particles from a visualized path line. PSV has a more reliable performance in particle matching as compared to particle tracking velocimetry and is therefore less likely to cause erroneous matching even in high-density images. The center of gravity of the first and last trajectories is obtained to calculate the displacement. The trajectory of the particle is illuminated using a diode laser and imaged using a digital single-lens reflex camera; the trajectory is then divided into three parts and recorded in a single frame using coded illumination. The first and second trajectories are short, and the third trajectory is long. The asymmetry of the trajectories is then used to determine the flow direction. We first evaluate the detection rate by increasing the trajectory density of synthetic images. The image size was fixed at 500 × 500 pixels, and the number of trajectories was increased from 28 to 280, and the detection rate was examined. Then, we evaluated the accuracy of detection of the center of gravity of the first and last trajectories using the root mean square error. Finally, we used the coded illumination method to visualize the swirling flow inside a device to examine its applicability to real flows.

Background: The spot, streak and rust are the most common diseases in maize, all of which require effective methods to recognize, diagnose and handle. This paper presents a novel image classification approach to the high accuracy recognition of these maize diseases. Methods: Firstly, the k-means clustering algorithm is deployed in LAB color space to reduce the influence of image noise and irrelevant background, so that the area of maize diseases could be effectively extracted. Then the statistic pattern recognition method and gray level co-occurrence matrix (GLCM) method are jointly used to segment the maize disease leaf images for accurately obtaining their texture, shape and color features. Finally, Support Vector Machine (SVM) classification method is used to identify three diseases. Results: Numerical results clearly demonstrate the feasibility and effectiveness of the proposed method. Conclusion: Our future work will focus on the investigation of how to use the new classification methods in dimensional and large scale data to improve the recognizing performance and how to use other supervised feature selection methods to improve the accuracy further.

Data-driven rain streak removal methods, which most of rely on synthesized paired data, usually come across the generalization problem when being applied in real cases. In this paper, we propose a novel deep-learning based rain streak removal method injected with self-supervision to improve the ability to remove rain streaks in various scales. To realize this goal, we made efforts in two aspects. First, considering that rain streak removal is highly correlated with texture characteristics, we create a fractal band learning (FBL) network based on frequency band recovery. It integrates commonly seen band feature operations with neural modules and effectively improves the capacity to capture discriminative features for deraining. Second, to further improve the generalization ability of FBL for rain streaks in various scales, we add cross-scale self-supervision to regularize the network training. The constraint forces the extracted features of inputs in different scales to be equivalent after rescaling. Therefore, FBL can offer similar responses based on solely image content without the interleave of scale and is capable to remove rain streaks in various scales. Extensive experiments in quantitative and qualitative evaluations demonstrate the superiority of our FBL for rain streak removal, especially for the real cases where very large rain streaks exist, and prove the effectiveness of its each component. Our code will be public available at: https://github.com/flyywh/AAAI-2020-FBL-SS.

A streak artifact is an anomaly seen in the CT image of the brain. It is like a dark bands or streaks appearing in an image that is not present in the original region. The appearance of these artifacts leads to misinterpretation of medical diagnosis. Under sampling, photon starvation, motion and beam hardening are some of the reasons for occurrence of streak artifacts. These type of artifacts commonly occur in the posterior fossa of the brain. This paper mainly focuses on region shape and texture identification and removal of streak artifacts exists in the brain image. Initially morphological operators and fuzzy k-means algorithms are used to identify the different regions in the CT brain image. Each region is treated as a single image and features are extracted for each image. The feature descriptors are constructed using region properties and statistical features. The extracted features are compared with threshold brain image to identify the dark bands of streak artifacts.

Single image de-raining is challenging especially in the scenarios with dense rain streaks. Existing methods resolve this problem by predicting the rain streaks of the image, which constrains the network to focus on local rain streaks features. However, dense rain streaks are visually similar to mist or fog (with large intensities), in this case, the training objective should be shifted to image recovery instead of extracting rain streaks. In this paper, we propose a coupled rain streak and background estimation network that explores the intrinsic relations between two tasks. In particular, our network produces task-dependent feature maps, each part of the features correspond to the estimation of rain streak and background. Furthermore, to inject element-wise attention to all the convolutional blocks for better understanding the rain streaks distribution, we propose a Separable Element-wise Attention mechanism. In this way, dense element-wise attention can be obtained by a sequence of channel and spatial attention modules, with negligible computation. Extensive experiments demonstrate that the proposed method outperforms state-of-the-arts on 5 existing synthesized rain datasets and the real-world scenarios, without extra multi-scale or recurrent structure.

Rain removal in videos is a problem that has attracted tremendous interest of researchers within the field of deep learning. Although deep-learning-based rain removal methods outperform large number of conventional vision methods, some technical issues that need to be resolved remain. In this article, we propose a new deep learning method for video rain removal based on recurrent neural network (RNN) architecture. Pseudo groundtruth was generated from real rainy video sequence by temporal filtering for supervised learning. Instead of focusing on various shapes of rain streaks similar to conventional methods, in this article, we focused on the changing behaviors of rain streaks. To accomplish this, images of progressive rain streaks were generated from the real rain videos and are sequentially fed to the network in a decreasing rain order. Multiple images with different amounts of rain streaks were used as RNN inputs to more efficiently identify rain streaks and then remove them. Experimental results demonstrate that our method is suitable for a wide range of rainy images. Moreover, experiments performed on both real-world and synthetic images demonstrate that our proposed method can achieve competitive results in comparison with the benchmarked and conventional approaches for rain streak removal from images.

Various weather conditions, such as rain, haze, or snow, can degrade visual quality in images/videos, which may significantly degrade the performance of related applications. In this paper, a novel framework based on sequential dual attention deep network is proposed for removing rain streaks (deraining) in a single image, called by SSDRNet (Sequential dual attention-based Single image DeRaining deep Network). Since the inherent correlation among rain steaks within an image should be stronger than that between the rain streaks and the background (non-rain) pixels, a two-stage learning strategy is implemented to better capture the distribution of rain streaks within a rainy image. The two-stage deep neural network primarily involves three blocks: residual dense blocks (RDBs), sequential dual attention blocks (SDABs), and multi-scale feature aggregation modules (MAMs), which are all delicately and specifically designed for rain removal. The two-stage strategy successfully learns very fine details of the rain steaks of the image and then clearly removes them. Extensive experimental results have shown that the proposed deep framework achieves the best performance on qualitative and quantitative metrics compared with state-of-the-art methods. The corresponding code and the trained model of the proposed SSDRNet have been available online at https://github.com/fityanul/SDAN-for-Rain-Removal .

With the rapid development of outdoor vision system, removing rain streaks from a single image has attracted considerable attention as rain streaks can affect the quality of the image taken in rainy days, and interrupt the key information, which will greatly reduce the use value of the image, thus affecting the performance of traffic, safety monitoring and other facilities. Although the deep learning methods have achieved satisfying performance in single image de-raining, there are still two problems: First, the rain streaks contained in one dataset we can use are limited, and in the case of real rainy days, the rain streak density is diverse, it is impossible to accurately classify them. Therefore, the existing rain removal models cannot remove rain streaks properly for images with different rain streak density which attend to over or under rain removal. Secondly, the results of single image after rain removal model often appear the phenomenon of variegated spots, image contrast saturation change and even unsmooth rain streak after rain removal. We use a three-way multi-scale weighted fusion module to enhance the feature extraction, and then generate an attention map through the improved spatial attentive module to accurately locate the location of the rain streaks. After the combination of the two, we will obtain the foreground information of the rain streaks. Through the characteristic of mutual game in the training mechanism of GAN, we can enhance the rain streak location recognition and effectively remove the rain at the same time. Through the training mechanism of the GAN network game, we can enhance the rain line location recognition and effectively remove the rain at the same time. Experiments show that our network achieves superior performance, it has high generalization for different rain streak density, and ensures that the contrast and saturation of the image are not changed.

Data-driven rain streak removal methods, most of which rely on synthesized paired data, usually come across the generalization problem when being applied in real scenarios. In this paper, we propose a novel deep-learning based rain streak removal method injected with self-supervision to obtain the capacity of removing more varied-scale rain streaks in practical applications. To this end, in this work, efforts are made from two perspectives. First, considering that rain streak removal is highly correlated with texture characteristics, we create a fractal band learning (FBL) network based on frequency band recovery. It integrates commonly seen band feature operations as neural forms and effectively improves the capacity to capture discriminative features for deraining. Second, to further improve the generalization ability of FBL to remove rain streaks of varied scales, we incorporate scale-robust self-supervision to regularize the network training. The constraint forces the extracted features of an input rain image at different scales to be equivalent after rescaling operations. Therefore, our method can offer similar responses based on solely image content without the interference of scale change and is capable to remove varied-scale rain streaks. Extensive experiments in quantitative and qualitative evaluations demonstrate the superiority of our method for rain streak removal, especially for the real cases where very large rain streaks exist, and prove the effectiveness of each component.

Although lowering X-ray radiation helps reduce the risk of cancer in patients, low-dose computed tomography (LDCT) technology usually leads to poor image quality, such as amplified mottle noise and streak artifacts, which severely impact the diagnostic results. To improve diagnostic performance, we propose an algorithm based on a cycle-consistent generative adversarial network (CycleGAN) to suppress noise and reduce artifacts. In addition, we include attention mechanisms in the proposed network to expand the receptive field and capture richer contextual dependencies. Unlike traditional methods that manually match similar local blocks, our proposed method can autonomously learn the relationship between local features and their global dependencies. Specifically, two different types of attention modules (criss-cross self-attention (CCSA) and channel attention (CA)) are adopted to enhance feature interdependencies in the spatial and channel dimensions separately. Because of the CCSA mechanism, noise and artifacts can be restored using cues from all local feature locations; the CA mechanism adaptively reassigns the weights of each feature map. Furthermore, we also performed a diagnostic quality assessment of the results and ablation studies of the loss functions and the structural modules, which showed the validity of our proposed method. Extensive experiments show that our proposed method achieves better metrics and visual effects than state-of-the-art methods.

Single image de-raining is an extremely challenging problem since the rainy images contain rain streaks which often vary in size, direction and density. This varying characteristic of rain streaks affect different parts of the image differently. Previous approaches have attempted to address this problem by leveraging some prior information to remove rain streaks from a single image. One of the major limitations of these approaches is that they do not consider the location information of rain drops in the image. The proposed Image Quality-based single image Deraining using Confidence measure (QuDeC), network addresses this issue by learning the quality or distortion level of each patch in the rainy image, and further processes this information to learn the rain content at different scales. In addition, we introduce a technique which guides the network to learn the network weights based on the confidence measure about the estimate of both quality at each location and residual rain streak information (residual map). Extensive experiments on synthetic and real datasets demonstrate that the proposed method achieves significant improvements over the recent state-of-the-art methods., TIP2019 submission. arXiv admin note: substantial text overlap with arXiv:1906.11129

This paper presents an algorithm to detect very faint object streaks on CCD images acquired with an optical system. The proposed detection method uses image filters simulating the geometrical form and orientation of possible streaks on the CCD image. The method searches for a matching between streak and filter evaluating the convolutions of the image with all possible filters. Based on the statistics of the image background a threshold is applied in order to accept, respectively reject, the possible streak candidates. The detection probabilities and the effect of different parameter settings are estimated with tests on simulated images, while subframes of real images are used to evaluate the applicability of the algorithm to real cases. The detection probability of this method depends on the length and on the signal-to-noise ratio of the streak. For long streaks, a detection for signal-to-noise values around 0.5 is achieved. The further characterization of the detected streak (e.g. centroid and length), which is not performed in the current algorithm, and the reduction of the computation time, which is relatively high for full acquired frames, as well as possible improvements are briefly addressed.

Efficiency and accuracy of semi-global matching (SGM) make it outperform many stereo matching algorithms and is widely used under challenging occasions. However, SGM only incorporates information along a scanline in each pass and lacks interaction between scanlines, resulting in streak artifacts in the disparity image. We introduce a local edge-aware filtering method to SGM to enhance the interaction of neighboring scanlines, since streak artifacts can be avoided. We use bilateral weights based on intensity similarity and spatial affinity between pixels to build connections among scanlines. In each pass, we recursively estimate the aggregated cost of SGM and compute the weighted average of aggregated costs for pixels in the orthogonal direction to obtain the output of our method along each scanline. As one-dimensional bilateral filtering is used in our method, the extra computation is linear to image resolution and label space, which is a small fraction of that needed by SGM. We present ablation studies using stereo pairs under both constrained and natural conditions to verify the effectiveness of our method. Extensive experiments on Middlebury and Karlsruhe Institute of Technology and Toyota Technology Institute datasets demonstrate that our method removes all streak artifacts, improves the quality of the disparity image, and outperforms many other non-local cost aggregation approaches.

In this paper, we present a disentangled representation learning and enhancement network (DRLE-Net) to address the challenging single image de-raining problems, i.e., raindrop and rain streak removal. Specifically, the DRLE-Net is formulated as a multi-task learning framework, and an elegant knowledge transfer strategy is designed to train the encoder of DRLE-Net to embed a rainy image into two separated latent spaces representing the task (clean image reconstruction in this paper) relevant and irrelevant variations respectively, such that only the essential task-relevant factors will be used by the decoder of DRLE-Net to generate high-quality de-raining results. Furthermore, visual attention information is modeled and fed into the disentangled representation learning network to enhance the task-relevant factor learning. To facilitate the optimization of the hierarchical network, a new adversarial loss formulation is proposed and used together with the reconstruction loss to train the proposed DRLE-Net. Extensive experiments are carried out for removing raindrops or rainstreaks from both synthetic and real rainy images, and DRLE-Net is demonstrated to produce significantly better results than state-of-the-art models.

In this paper, we address the rain streak removal from a single image. In order to efficiently detect and remove the annoying rain streaks, we propose a global single-directional gradient prior with the L0 norm to model the rain streak. To preserve the abundant information of the background, we learn a convolutional sparse coding (CSC) to represent the background. Furthermore, we develop an alternating direction method of multipliers (ADMM) to solve multi-variable optimization problems. Experiments on synthesized and real-world images show that the proposed method outperforms state-of-art methods in terms of rain streak removal and background preservation.

Compressed ultrafast photography (CUP) is the fastest receive-only single-shot imaging technique up to now. By combining compressed sensing and streak imaging, CUP is able to capture ultrafast dynamics in a single shot. As a powerful tool for researching ultrafast phenomena, it has been widely applied in lots of areas. To meet the demand for more precise dynamics information and higher dimension in some application, many improvements have been conduct in CUP. For example, we have raised total variation-block match 3D filter algorithm and augmented Lagrange-deep learning hybrid algorithm to improve the reconstructed image quality of CUP, and set up a stereo-volumetric CUP system to capture 5 dimension dynamic information in a single shot. Besides, we have also developed another single-shot ultrafast optical imaging technique, chirped spectral mapping ultrafast photography (CSMUP), which utilized the spectral-temporal mapping to exact temporal information from hyperspectral image.

The Streaking effect is an artifact or characteristic fingerprint left by the application of median filter. Many studies have used the application of median filters on digital images. Recently, percentage streaking area (PSA) has been used as a metric to measure streaking in images. The paper performs an analysis of the percentage streak area (PSA). The work investigates streaking effect using PSA in natural images and when commonly applied filters such as Gaussian filter, Average filter, and Unsharp masking are used for image filtering. Standard image datasets UCID, BOSS, and Dresden have been used in the study, and results are presented. The investigation shows that the streaking effect can be quantified by the percentage streak area.

Although supervised single image deraining (SID) have obtained impressive results, they still cannot obtain satisfactory results on real images for the weak generalization of rain removal capacity. In this paper, we mainly discuss the semi-supervised SID and propose a new GAN-based deraining network called Semi-DerainGAN, which can use both synthetic and real data in a uniform network based on two supervised and unsupervised processes. For this task, a semi-supervised rain streak learner termed SSRML sharing the same parameters of both processes is derived, which makes the real images contribute more rain streak information, so that the resulted model has a strong generalization power to the real SID task. We also contribute a new real-world rain image dataset called Real200 to alleviate the difference between both synthetic and real image domains. Extensive results on public datasets show that our model can obtain competitive results, especially on the real rain images.

Rain produces haze and streak noise in an image that destroys the intricate details, turning even the most effective computer vision algorithms obsolete. This calls for a good deraining algorithm to restore a rainy image to a rain-less image with natural sharpness, contrast and brightness. In this paper, we propose SphinxNet with a less number of parameters by utilizing autoencoders in an intelligent formation to derive maximum spatial awareness. Which helps it to achieve not only sharp but better derained results than other models with more number of trainable parameters. Autoencoders are made of linear sequence of convolutional layers with channel wise bottleneck and skip concatenation connections between encoder and decoder layers. We trained our network on popular synthetic datasets. In quantitative and qualitative analysis we found our method to perform not only better but more consistent on variety of images.

Existing rain-removal algorithms often tackle either rain streak removal or raindrop removal, and thus may fail to handle real-world rainy scenes. Besides, the lack of real-world deraining datasets comprising different types of rain and their corresponding rain-free ground-truth also impedes deraining algorithm development. In this paper, we aim to address real-world deraining problems from two aspects. First, we propose a complementary cascaded network architecture, namely CCN, to remove rain streaks and raindrops in a unified framework. Specifically, our CCN removes raindrops and rain streaks in a complementary fashion, i.e., raindrop removal followed by rain streak removal and vice versa, and then fuses the results via an attention based fusion module. Considering significant shape and structure differences between rain streaks and raindrops, it is difficult to manually design a sophisticated network to remove them effectively. Thus, we employ neural architecture search to adaptively find optimal architectures within our specified deraining search space. Second, we present a new real-world rain dataset, namely RainDS, to prosper the development of deraining algorithms in practical scenarios. RainDS consists of rain images in different types and their corresponding rain-free ground-truth, including rain streak only, raindrop only, and both of them. Extensive experimental results on both existing benchmarks and RainDS demonstrate that our method outperforms the state-of-the-art.

In this paper, we address the problem of rain streaks and rain accumulation removal in video, by developing a self-alignment network with transmission-depth consistency. Existing video based deraining methods focus only on rain streak removal, and commonly use optical flow to align the rain video frames. However, besides rain streaks, rain accummulation can considerably degrade visibility; and, optical flow estimation in a rain video is still erroneous, making the deraining performance tend to be inaccurate. Our method employs deformable convolution layers in our encoder to achieve feature-level frame alignment, and hence avoids using optical flow. For rain streaks, our method predicts the current frame from its adjacent frames, such that rain streaks that appear randomly in the temporal domain can be removed. For rain accumulation, our method employs a transmission-depth consistency loss to resolve the ambiguity between the depth and water-droplet density. Our network estimates the depth from consecutive rain-accumulation-removal outputs, and calculates the transmission map using a commonly used physics model. To ensure photometric-temporal and depth-temporal consistencies, our method estimates the camera poses, so that it can warp one frame to its adjacent frames. Experimental results show that our method is effective in removing both rain streaks and rain accumulation, outperforming those of state-of-the-art methods quantitatively and qualitatively.

Rain streaks bring serious blurring and visual quality degradation, which often vary in size, direction and density. Current CNN-based methods achieve encouraging performance, while are limited to depict rain characteristics and recover image details in the poor visibility environment. To address these issues, we present a Multi-scale Hourglass Hierarchical Fusion Network (MH2F-Net) in end-to-end manner, to exactly captures rain streak features with multi-scale extraction, hierarchical distillation and information aggregation. For better extracting the features, a novel Multi-scale Hourglass Extraction Block (MHEB) is proposed to get local and global features across different scales through down- and up-sample process. Besides, a Hierarchical Attentive Distillation Block (HADB) then employs the dual attention feature responses to adaptively recalibrate the hierarchical features and eliminate the redundant ones. Further, we introduce a Residual Projected Feature Fusion (RPFF) strategy to progressively discriminate feature learning and aggregate different features instead of directly concatenating or adding. Extensive experiments on both synthetic and real rainy datasets demonstrate the effectiveness of the designed MH2F-Net by comparing with recent state-of-the-art deraining algorithms. Our source code will be available on the GitHub: https://github.com/cxtalk/MH2F-Net., Accepted in CVPRW 2021

We discuss the use of matrix relevance learning, a popular extension to prototype learning algorithms, applied to a three-class classification task of diagnosing cassava diseases from spectral data. Previously this diagnosis has been done using plant image data taken with a smartphone. However for this method disease symptoms need to be visible. Unfortunately for some cassava diseases, once symptoms have manifested on the aerial part of the plant, the root which is the edible part of the plant has been totally destroyed. This research is premised on the hypothesis that diseased crops without visible symptoms can be detected using spectral information, allowing for early interventions. In this paper, we analyze visible and near-infrared spectra captured from leaves infected with two common cassava diseases (cassava brown streak disease and cassava mosaic virus disease) found in Sub-Saharan Africa. We also take spectra from leaves of healthy plants. The spectral data come with thousands of dimensions, therefore different wavelengths are analyzed in order to identify the most relevant spectral bands for diagnosing these disease. To cope with the nominally high number of input dimensions of data, functional decomposition of the spectra is applied. The classification task is addressed using Generalized Matrix Relevance Learning Vector Quantization and compared with the standard classification techniques performed in the space of expansion coefficients.

In radiation treatment, a cone-beam computed tomography (CBCT) scan is conducted for precise positioning of tumors, and the image quality is usually degraded by motion artifacts due to patient’s respiration and movement during scanning. Four-dimensional (4D) CBCT reconstruction with phase binning is typically used to overcome these difficulties. Albeit motion artifacts might be reduced with 4D CBCT, the overall image quality is typically worsened by severe streak artifacts due to the sparse-angle projections available in the 3D reconstruction for each motion phase. This study presents a method for reducing streak artifacts effectively in conventional 4D CBCT reconstruction by using a state-of-the-art convolutional neural network (a residual U-Net was used). We performed a computational simulation and an experiment to investigate the image quality and evaluate the effectiveness of the proposed method. The proposed 4D CBCT reconstruction method reduced streak artifacts noticeably, and its effectiveness was validated by comparing its results to those of other reconstruction methods such as the filtered-backprojection, a compressed-sensing, and a simple CNN-based algorithm for the 4D CBCT datasets.

The aim of this paper is to develop a fast and accurate beam hardening correction method by modeling physical interactions between X-ray photons and materials for computed tomography (CT) imaging. The nonlinear attenuation process of the X-ray projection is modeled by reprojecting a template image with the estimated polychromatic spectrum. By adding the scaled difference of the monochromatic reprojection data and the polychromatic reprojection to the raw projection data, the raw projection data is mapped into the corresponding monochromatic projection data, which is to reconstruct the beam hardening artifacts corrected images. The algorithm can also be implemented in image-domain which takes the uncorrected image volume as input when there is an adequate model of the spectrum. In this case, the scaled difference is reconstructed to yield a set of artifacts images that can be added directly to the uncorrected images. Numerical simulations, experimental phantom data, and animal data which are acquired on a modern diagnostic CT scanner (Discovery CT750 HD, GE Healthcare, WI, USA), and a modern C-Arm CT scanner (Artis Zee, Siemens Healthcare, Forchheim, Germany), respectively, are used to evaluate the proposed method. The results show the proposed method significantly reduced both cupping and streak artifacts, and successfully recovered the Housfield units accuracy. Extensive studies suggest the proposed model-based method successfully corrects the beam hardening artifacts. This paper is practically useful and is promising to be applied to commercial products.

In this paper, we address a rain removal problem from a single image, even in the presence of large rain streaks and rain streak accumulation (where individual streaks cannot be seen and thus are visually similar to mist or fog). For rain streak removal, the mismatch problem between different streak sizes in training and testing phases leads to poor performance, especially when there are large streaks. To mitigate this problem, we embed a hierarchical representation of wavelet transform into a recurrent rain removal process: 1) rain removal on the low-frequency component and 2) recurrent detail recovery on high-frequency components under the guidance of the recovered low-frequency component. Benefiting from the recurrent multi-scale modeling of wavelet transform-like design, the proposed network trained on streaks with one size can adapt to those with larger sizes, which significantly favors real rain streak removal. The dilated residual dense network is used as the basic model of the recurrent recovery process. The network includes multiple paths with different receptive fields, thus it can make full use of multi-scale redundancy and utilize context information in large regions. Furthermore, to handle heavy rain cases where rain streak accumulation is presented, we construct a detail appearing rain accumulation removal to not only improve the visibility but also enhance the details in dark regions. The evaluation of both synthetic and real images, particularly on those containing large rain streaks and heavy accumulation, shows the effectiveness of our novel models, which significantly outperforms the state-of-the-art methods.

We present DeepStreaks, a convolutional-neural-network, deep-learning system designed to efficiently identify streaking fast-moving near-Earth objects that are detected in the data of the Zwicky Transient Facility (ZTF), a wide-field, time-domain survey using a dedicated 47 deg2 camera attached to the Samuel Oschin 48-inch Telescope at the Palomar Observatory in California, United States. The system demonstrates a 96–98 per cent true positive rate, depending on the night, while keeping the false positive rate below 1 per cent. The sensitivity of DeepStreaks is quantified by the performance on the test data sets as well as using known near-Earth objects observed by ZTF. The system is deployed and adapted for usage within the ZTF Solar system framework and has significantly reduced human involvement in the streak identification process, from several hours to typically under 10 min per day.

Background and purpose: Computed tomography (CT) radiomics of head and neck cancer (HNC) images is susceptible to dental implant artifacts. This work devised and validated an automated algorithm to detect CT metal artifacts and investigate their impact on subsequent radiomics analyses. A new method based on features from total variation, gradient directional distribution, and Hough transform was developed and evaluated. Materials and methods: Two HNC datasets were analyzed: a training set of 131 patients for developing the detection algorithm and a testing set of 220 patients. Seven designated features were extracted from ROIs (regions of interest) and machine learning with random forests was used for building the artifact detection algorithm. Performance was assessed using the area under the receiver operating characteristics curve (AUC). Results: The testing results of artifacts detection yielded a cross-validated AUC of 0.91 (95% CI: 0.89–0.94), and a test AUC of 0.89. External testing validation yielded an accuracy of 0.82. For radiomics model prediction, training with artifacts yielded an AUC of 0.64 (95% CI: 0.63–0.65), while training on images without artifacts improved the AUC to 0.75 (95% CI: 0.74–0.76). This was compared to visual inspection of artifacts (AUC = 0.71 [95% CI: 0.69–0.73]). Conclusion: We developed a new method for automated and efficient detection of streak artifacts. We also showed that such streak artifacts in HNC CT images can worsen the performance of radiomics modeling. Keywords: Artifact detection, Radiomics, Machine learning

In dental computed tomography (DCT), metal artifact reduction (MAR) is a critical issue to improve the clinical usefulness of DCT and remains a challenging problem. Although various MAR methods have been developed in medical CT, those methods may not work robustly in DCT because teeth themselves, as well as metallic objects, have high X-ray attenuation. In this study, we investigated an MAR method that was based on sinogram normalization interpolation with an artifact-reduced prior for DCT. The method consisted of three main steps: segmentation of a metal trace, generation of an artifact-reduced prior image, and sinogram completion followed by DCT reconstruction. We performed a computational simulation and performed an experiment on a teeth phantom with several metal inserts to validate the proposed method. With respect to the root-mean-square error and the structural similarity, we compared our results with the ones obtained by using the combined prior-based MAR (CP-MAR) method. Our results indicate that the proposed MAR method reduced metal artifacts considerably in DCT images and showed an image performance that was better than that obtained by using the state of the art method (CP-MAR) in reducing streak artifacts without introducing any contrast anomaly.

Long exposure time and wide field can effectively improve the ability of a space surveillance telescope to detect faint space targets. However, complicated situations pose challenges for space target detection. Background star images usually manifest a rotated streak, and target trajectories can be crossed, discontinuous, or nonlinear. This paper presents an accurate and robust space target detection method, namely, spatiotemporal pipeline multistage hypothesis testing (SPMHT), to overcome the issues. Specifically, the method includes the following two stages: First, in the spatiotemporal pipeline filtering step, Spatiotemporal-related Intersection over Union (SrIoU) is used to calculate the IoU score instead of the traditional method. Benefiting from the differences between motion characteristics of targets and stars and the insensitivity of the SrIoU score to the noise, the spatiotemporal pipeline filtering can effectively remove the streak images of background stars and obtain candidate points. Second, a series of candidate points is further organized into a tree structure. We pruned in the tree structure combined with these candidate trajectories by using velocity and direction feature of moving objects. Furthermore, in the search step, fast adaptive sequence region search is used to reduce the computational cost. The experimental results for two datasets, simulated image datasets and real captured image datasets, demonstrate the effectiveness in addressing the difficulties of space target detection in complicated situations.

Computed tomography (CT) images with a low-dose protocol generally have severe mottle noise and streak artifacts. In this paper, we propose a novel diffusion method named “artifact suppressed nonlinear diffusion filtering (ASNDF),” to process low-dose CT (LDCT) images. Different from other diffusion filtering methods, the proposed ASNDF not only includes image gradient as the main cue to construct a diffusion coefficient function, but also incorporates the local variances of image to be diffused and residual image between two adjacent diffusions. In detail, the classical PM diffusion is first performed to get the initial residual image, and then from the second iteration, the LDCT image is processed according to the ASNDF processing. Simulated data, clinical data and rat data are conducted to evaluate the proposed method, and the comparison experiments with other competing methods show that the proposed ASNDF method makes an improvement in artifact suppression and structure preservation, and offers a sound alternative to process LDCT images from most current CT systems.

Low-dose computed tomography (LDCT) technique is an important imaging modality, but LDCT images are always severely degraded by mottle noise and streak artifacts. The recently proposed nonlocally centralized sparse representation (NCSR) algorithm has good performance in natural image denoising, but it suffers from residual streak artifacts and can't preserve edges structure information well when implemented in LDCT image denoising. In addition, it has high computational complexity. To address this problem, in this paper, we propose an improved model, i.e. SNCSR model, based on the stationary PCA sub-dictionaries, nonlocally centralized sparse representation and relative total variation. In the SNCSR model, in order to learn more accurate sub-dictionaries, the LDCT image is preprocessed by the improved total variation (ITV) model in which the weighted coefficient of the regularization term is constructed depending on a clipped and normalized local activity. In addition, the maximum eigenvalue of the gradient covariance matrix of the image patch is used to distinguish edge structure information from background region so that the restored image can be represented more sparsely. Moreover, unlike the NCSR model that needs to learn sub-dictionaries in each outer loop, the proposed model learns stationary sub-dictionaries only once before iteration starts, which shorten the computation time significantly. At last, the relative total variation (RTV) algorithm is applied to further reduce the residual artifacts in the recovered image more thoroughly. The experiments are performed on the simulated pelvis phantom, the actual thoracic phantom and the clinical abdominal data. Compared with several other competitive denoising algorithms, both subjective visual effect and objective evaluation criteria show that the proposed SNCSR model has lower computational complexity and can improve LDCT images quality more effectively.

Comprehensive two-dimensional gas chromatography (GC × GC) can separate thousands of different compounds, and is used for many important applications such as petrochemical processing and environmental monitoring, etc. GC × GC generates rich and complex information, which requires automated processing for rapid chemical identification and classification. A challenge is to remove unwanted streaks that may affect the quantification and identification of analytes. It is difficult to detect streaks because of complex backgrounds, low-contrast data, and variable shapes, scales, and orientations of streaks in GC × GC data. This paper proposes a new approach to detect streaks effectively based on image analysis techniques. By adopting a pseudo-log function and preprocessing methods to compress the original data and enhance the low-contrast data, we employ steerable Gaussian filtering to delineate streak regions based on the specific orientations of streaks. A marker-controlled watershed algorithm is then used to segment the streaks, and highly discriminating characteristics are used to identify candidate regions and reject false streaks. In the end, with a diverse data set generated from gas chromatograph, experiments are carried out and the results demonstrate that our streak detection approach is effective and robust with respect to changes in streak patterns, even in variable chromatographic conditions. The proposed object detection method effective in complex backgrounds and low-contrast conditions is also helpful for object detection in other scenes.

Rainfall weather always degrades the quality of the images severely in the outdoor surveillance system. To improve the quality of images, different rain removal algorithms have been proposed recently. As the decomposition based methods do not need to impose any restrictions on the types of rain, they have a wider application prospect. However, they still have the problems of rain residue in the low-frequency components and information loss in high-frequency components. To solve these problems, we propose an image rain streak removal algorithm based on the depth of field (DoF) and sparse coding. Firstly, we decompose the image by using the combination of bilateral filtering and short-time Fourier transform, so that the contour in the low-frequency part of the image can be better preserved. Then the DoF saliency map of the image is used both to reduce the rain residue in the low-frequency components and to avoid mis-matching the background and the rain streaks with the same gradient in the high-frequency components. We use DoF saliency map as the weights for the weighted sum of the original image and the initial low-frequency image to obtain the corrected low-frequency images. The DoF saliency map is also used to twice weaken the rain streaks in the high-frequency image to generate the corrected high-frequency images. The algorithm includes four steps: image decomposition, dictionary learning, atomic clustering based on Principal Component Analysis and Support Vector Machine, image revising based on DoF saliency map. The experimental results demonstrate that our proposed algorithm performs better both in rain removal and high-frequency information preserving than current methods.

To compare a fat-suppressed T2-weighted periodically rotated overlapping parallel lines with enhanced reconstruction (T2W-PROPELLER) sequence with a fat-suppressed T2-weighted fast spin-echo (T2W-FSE) sequence in the oral and maxillofacial regions for the evaluation of the presence of motion artifacts caused by mandibular movements.Fifty-six healthy adult volunteers were examined in a closed mouth position and then with three different rhythmical mandibular movements throughout MR scanning: open-close movement (movement 1), lateral movement (movement 2) and open-close and lateral movement (movement 3). All subjects were scanned first with fat-suppressed T2W-FSE and then with fat-suppressed T2W-PROPELLER while performing the same movements. Motion artifacts, including ghosting or pulsation artifacts, streak artifacts, susceptibility artifacts and the overall image quality were independently evaluated by two oral and maxillofacial radiologists using a five-point scale. The score graded by the two observers was averaged.The inter-observer agreement was almost perfect for all evaluated items (κ ≥ 0.81). The T2W-PROPELLER images showed significantly fewer ghosting artifacts than T2W-FSE images in subjects performing the mandibular movements throughout MR scanning (P .001). T2W-PROPELLER images also showed significantly fewer pulsation artifacts than T2W-FSE images, regardless of the performance of a movement, throughout MR scanning (P .001). Finally, the T2W-PROPELLER images showed a significantly better overall image quality than T2W-FSE images in subjects performing movements 2 or 3 throughout MR scanning (P .001).The PROPELLER technique was found to be effective in reducing the motion artifacts caused by mandibular movements on fat-suppressed T2W MR images in the oral and maxillofacial regions.

Single image deraining is regarded as an important research direction in image processing. To tackle the over-smoothing effect caused by the overlapping between rain streaks and the background, we propose a multi-fusion network for single image deraining. A novel local feature fusion block and a global feature fusion block are explored to fuse the high-level features with the low-level ones and correct the low-level representations. By stacking multiple fusion blocks, the proposed network can fully utilize the high-level information and extract powerful feature maps of rain streak layers. In addition, based on the prediction difficulty, a curriculum learning strategy is further explored to make the training process easier. Extensive experiments demonstrate that our network performs favorably against other deraining approaches.

This paper presents a machine learning approach using Neural Networks for the detection and localization of a single streaking target from an optical sensor's Focal Plane Array (FPA). Current state-of-the-art methods utilize model-based probabilistic techniques, such as the maximum likelihood method. The image data for training and the concomitant ground truth for detection and localization is generated via simulation due to lack of sufficient amounts of real-world data. The images were generated assuming that the target's point spread function (PSF) is Gaussian and moves during the FPA's integration time. Also, a noise model for each optical pixel on the FPA is used that is consistent with a Poisson model of the number of non-target originated photons. Our approach divides the problem into two parts: streak detection and localization. To simplify neural network architecture selection and training, we split the large images into smaller sub-images. The noisy sub-images were preprocessed by an autoencoder with Convolutional layers, which preserved the target intensity. Next, we used the denoised sub-images from the autoencoder to detect streaks. For localization, the Neural Network was given the original sub-images to predict the pixel locations of a streak's start and end points. All three processing steps utilized Convolutional layers. The performance of the detection model was evaluated in terms of confusion matrices at different noise intensities, as well as the Receiver Operating Characteristic (ROC) curves at a constant noise intensity. For the localization model, we evaluated the Mean Squared Error (MSE). These results were compared to the Generalized Likelihood Ratio Test (GLRT) for detection and an existing matched filter-based maximum likelihood method for localization. We found that the machine learning models were 340 times faster for detection and approximately 360 times faster for localization on full size 256x256 images. In addition, the detection accuracy of the machine learning model was substantially better than the maximum likelihood method (area under the ROC of 0.94 versus 0.85) and the MSE was significantly lower (0.0261 versus 0.0622 on 32x32 sub-images and 0.0351 versus 0.243 on 256x256 original images). These results are significant for time-critical systems, such as missile defense systems (e.g., C-RAM). Aside from missile defense, the approach presented here can be applied to counter many other weapons that exhibit similar linear streak trajectories, such as torpedoes. In this case, sonar images would be used rather than images from an optical sensor.

In this study, we propose a method to reduce streak artifacts of sparse view CT images via convolutional neural network (CNN). The main idea of the proposed method is to utilize both image and sinogram domain data for CNN training. To generate datasets, projection data were acquired from 512 (128) views using Siddon’s ray-driven algorithm, and full (sparse) view CT images were reconstructed by filtered back projection with a Ram-Lak filter. We first trained U-net based CNN_img, which was designed to reduce the streak artifacts of sparse view CT in image domain. Then, the output images of CNN_img were used as prior images to conduct pseudo full view sinogram. Before upsampling, sparse view sinogram was normalized by the prior images, and then linear interpolation was employed to estimate the missing view data compared to full view sinogram. The upsampled data were denormalized using prior images. To reduce the residual errors in pseudo full view sinogram data, we trained CNN_hybrid with residual encoder-decoder CNN, which is known to be effective in reducing the residual errors while preserving structural details. In order to increase the learning efficiency, the dynamic range of the pseudo full view sinogram data was converted via exponential function. The results show that the CNN_hybrid provides better performance in streak artifacts reduction than CNN_img, which is also confirmed by quantitative assessment.