This paper proposes an adaptive color-guided autoregressive (AR) model for high quality depth recovery from low measurements captured by cameras. We observe and verify that the AR tightly fits maps of generic scenes. The task is formulated into a minimization prediction errors subject to measurement consistency. predictor each pixel constructed according both local correlation in initial map nonlocal similarity accompanied color image. analyze stability our method linear system point view,...

Stereoscopic perception is an important part of human visual system that allows the brain to perceive depth. However, depth information has not been well explored in existing saliency detection models. In this letter, a novel method for stereoscopic images proposed. First, we propose measure evaluate reliability map, and use it reduce influence poor map on detection. Then, input image represented as graph, introduced into graph construction. After that, new definition compactness using color...

We propose a new generalised spatial modulation (GSM) technique, which can be considered as generalisation of the recently proposed (SM) technique. SM seen special case GSM with only one active transmit antenna. In contrast to SM, uses indices multiple antennas map information bits, and is thus able achieve substantially increased spectral efficiency. Furthermore, selecting enables harvest significant diversity gains in comparison because all same information. On other hand, inter-channel...

Co-saliency detection aims at extracting the common salient regions from an image group containing two or more relevant images. It is a newly emerging topic in computer vision community. Different most existing co-saliency methods focusing on RGB images, this paper proposes novel model for RGBD which utilizes depth information to enhance identification of co-saliency. First, intra saliency map each generated by single model, while inter calculated based multi-constraint feature matching,...

As a newly emerging and significant topic in computer vision community, co-saliency detection aims at discovering the common salient objects multiple related images. The existing methods often generate map through direct forward pipeline which is based on designed cues or initialization, but lack refinement-cycle scheme. Moreover, they mainly focus RGB image ignore depth information for RGBD In this paper, we propose an iterative framework, utilizes single saliency maps as generates final by...

In this paper, we propose to introduce intrinsic image decomposition priors into models for contrast enhancement. Since is a highly illposed problem, constraints on both reflectance and illumination layers yield reliable solution. We regularize the layer be piecewise constant by introducing weighted ℓ1 norm constraint neighboring pixels according color similarity, so that decomposed would not affected much information. The regularized smoothness constraint. proposed model effectively solved...

Separation of video clips into foreground and background components is a useful important technique, making recognition, classification, scene analysis more efficient. In this paper, we propose motion-assisted matrix restoration (MAMR) model for foreground-background separation in clips. the proposed MAMR model, backgrounds across frames are modeled by low-rank matrix, while objects sparse matrix. To facilitate efficient separation, dense motion field estimated each frame, mapped weighting...

Cloud detection is a crucial preprocessing step for optical satellite remote sensing (RS) images. This article focuses on the cloud RS imagery with cloud-snow coexistence and utilization of thumbnails that lose considerable amount high resolution spectrum information original images to extract mask efficiently. To tackle this problem, we propose novel neural network an encoder-decoder structure, named CDnetV2, as series work detection. Compared our previous CDnetV1, CDnetV2 contains two...

Video saliency detection aims to continuously discover the motion-related salient objects from video sequences. Since it needs consider spatial and temporal constraints jointly, is more challenging than image detection. In this paper, we propose a new method detect in based on sparse reconstruction propagation. With assistance of novel static motion priors, single-frame model first designed represent each individual frame via sparsity-based reconstruction. Then, through progressive...

The 3D extension of High Efficiency Video Coding significantly improves the coding efficiency video at expense computational complexity. This paper presents a novel fast mode decision algorithm for depth map based on grayscale similarity and inter-view correlation. First, is adopted to judge whether reference frame could assist current frame. When difference in average between co-located unit (CU) CU smaller than threshold, level will be restricted by that coded CU. Second, correlation are...

Multispectral remote sensing (RS) images are often contaminated by the haze that degrades quality of RS data and reduces accuracy interpretation classification. Recently, emerging deep convolutional neural networks (CNNs) provide us new approaches for image dehazing. Unfortunately, power CNNs is limited lack sufficient hazy-clean pairs imagery, which makes supervised learning impractical. To meet hunger CNNs, we propose a novel synthesis method to generate realistic hazy multispectral...

Occluded person re-identification (re-id) aims to match occluded images holistic ones. Most existing works focus on matching collective-visible body parts by discarding the parts. However, only preserving causes great semantic loss for images, decreasing confidence of feature matching. On other hand, we observe that can provide missing information same identity. Thus, compensating image with its counterpart has potential alleviating above limitation. In this paper, propose a novel Reasoning...

We present an effective method for defocus map estimation from a single natural image. It is inspired by the observation that defocusing can significantly affect spectrum amplitude at object edge locations in By establishing relationship between amount of spatially varying blur and contrast locations, we first estimate these then full be obtained propagating over entire image with nonhomogeneous optimization procedure. The proposed takes into consideration not only light refraction but also...

In this paper, we propose a novel universal framework for salient object detection, which aims to enhance the performance of any existing saliency detection method. First, rough regions are extracted from model with distance weighting, adaptive binarization, and morphological closing. With superpixel segmentation, Bayesian decision is adopted refine map obtain more accurate map. An iterative optimization method designed better results by exploiting characteristics output each time. Through...

Screen content coding (SCC) has evolved into the extension of High Efficiency Video Coding (HEVC). Low-latency, real-time transport between devices in form screen video is becoming popular many applications. However, complexity encoder still very high for intra prediction HEVC-based SCC. This paper proposes a fast method based on property analysis First, units (CUs) are classified natural CUs (NCCUs) and (SCCUs), statistic characteristics content. For NCCUs, newly adopted modes, including...

In this paper, the two-dimensional (2-D) direction-of-arrival (DOA) estimation problem for a mixture of circular and noncircular sources is considered. particular, we focus on 2-D array structure consisting two parallel uniform linear arrays build general model with mixed sources. The received data its conjugate counterparts are combined together to form new vector, based which series DOA estimators derived. Compared existing methods, proposed one has three main advantages. First, it can...

Depth-Image-Based-Rendering (DIBR) techniques are significant for three-dimensional (3D) video applications, e.g., 3D television and free viewpoint (FVV). Unfortunately, the DIBR-synthesized image suffers from various distortions, which induce an annoying viewing experience entire FVV. Proposing a quality evaluator images is fundamental design of perceptual friendly FVV systems. Since associated reference usually not accessible, full-reference (FR) methods cannot be directly applied...

As a typical task of passive biometrics, behavior-based person identification has been studied extensively in recent years. This letter proposes the use ultrawideband impulse radar for based upon micro-Doppler signatures human motions. A new convolutional neural network (CNN) architecture is proposed taking advantage hierarchical features. The experimental results show that, by utilizing six selected motions, can be accurately achieved. Both traditional algorithms and landmark CNN are chosen...

In remote sensing, micro-Doppler signatures are widely used in moving target detection and automatic recognition. However, since Doppler easily affected by the direction of target, prior information aspect angle is essential for spectral analysis. Thus, a micro-Doppler-based classifier considered to be "angle-sensitive." this article, we propose an angle-insensitive omnidirectional classification problem using monostatic radar through proposed new convolutional neural network. We further...

Currently, many tone mapping operators (TMOs) have been provided to compress high dynamic range images low (LDR) for visualizing them on the common displays. Since quality degradation is inevitably induced by compression, how evaluate obtained LDR indeed a headache problem. Until now, only few full reference (FR) image assessment metrics proposed. However, they highly depend and neglect human visual system characteristics, hindering practical applications. In this paper, we propose an...

Nowadays, high-dynamic-range (HDR) imaging represents a prevailing trend and attracts much attention from both academic industrial scholars. Since HDR images cannot be properly produced on the mainstream low-dynamic-range (LDR) displays, various tone-mapping operators or postprocessing technologies have been designed to transform into LDR for visualization displays. However, it inevitably induces artifacts distortions due dynamic range compression. Besides, existing tone-mapped (TM)...