Pixel Domain Research Articles

Blind correction of the EB-leakage in the pixel domain

We study the problem of EB-leakage that is associated with incomplete polarized CMB sky. In the blind case that assumes no additional information about the statistical properties and amplitudes of the signal from the missing sky region, we prove that the recycling method (Liu et al. 2018) gives the unique best estimate of the EB-leakage. Compared to the previous method, our method reduces the uncertainties in the BB power spectrum due to EB-leakage by more than one order of magnitude in the most interesting domain of multipoles, where ℓ is between 100 and 200. This work also provides a useful guideline for observational design of future CMB experiments.

Survey of visual just noticeable difference estimation

The concept of just noticeable difference (JND), which accounts for the visibility threshold (visual redundancy) of the human visual system, is useful in perception-oriented signal processing systems. In this work, we present a comprehensive review of JND estimation technology. First, the visual mechanism and its corresponding computational modules are illustrated. These include luminance adaptation, contrast masking, pattern masking, and the contrast sensitivity function. Next, the existing pixel domain and subband domain JND models are presented and analyzed. Finally, the challenges associated with JND estimation are discussed.

A Natural Steganography Embedding Scheme Dedicated to Color Sensors in the JPEG Domain

Using Natural Steganography (NS), a cover raw image acquired at sensitivity ISO 1 is transformed into a stego image whose statistical distribution is similar to a cover image acquired at sensitivity ISO 2 > ISO 1. This paper proposes such an embedding scheme for color sensors in the JPEG domain, extending thus the prior art proposed for the pixel domain and the JPEG domain for monochrome sensors. We first show that color sensors generate strong intra-block and inter-block dependencies between DCT coefficients and that theses dependencies are due to the demosaicking step in the development process. Capturing theses dependencies using an empirical covariance matrix, we propose a pseudo-embedding algorithm on greyscale JPEG images which uses up to four sub-lattices and 64 lattices to embed information while preserving the estimated correlations among DCT coefficients. We then compute an approximation of the average embedding rate w.r.t. the JPEG quality factor and evaluate the empirical security of the proposed scheme for linear and non-linear demosaicing schemes. Our experiments show that we can achieve high capacity (around 2 bit per nzAC) with a high empirical security (P E 30% using DCTR at QF 95).

Visual Attention Guided Pixel-Wise Just Noticeable Difference Model

The just noticeable difference (JND) models in pixel domain are generally composed of luminance adaptation (LA) and contrast masking (CM), which takes edge masking (EM) and texture masking (TM) into consideration. However, in existing pixel-wise JND models, CM is not evaluated appropriately since they overestimate the masking effect of regular oriented texture regions and neglect the visual attention characteristic of human eyes for the real image. In this work, a novel JND model in pixel domain is proposed, where orderly texture masking (OTM) for regular texture areas (also called orderly texture regions) and disorderly texture masking (DTM) for complex texture areas (also called disorderly texture regions) are presented based on the orientation complexity. Meanwhile, the visual saliency is set as the weighting factor and is incorporated into CM evaluation to enhance JND thresholds. Experimental results indicate that compared with existing relevant JND profiles, the proposed JND model tolerates more distortion in the same perceptual quality, and brings better visual perception in the same level of the injected JND-noise energy.

HEVC/H.265 Intra Coding Based on the Human Visual System

High efficiency video coding (HEVC) is standard in the video compression field and performs well not only in video compression coding, but also in image compression. Regions of interest (ROIs) and just noticeable difference (JND), two human visual models, can accurately quantify human visual system (HVS) characteristics as pixel values. Using ROIs and JND to assist in evaluating image distortion can effectively reduce human visual redundancy and reflect authentic perceptual distortions. However, they are not readily applicable to the HEVC test model (HM) in the pixel domain. It is difficult to secure a suitable Lagrange multiplier $\lambda $ and quantization parameter (QP) for the JND model in particular. This paper proposes different solutions for the use of rate control (RC) or not where an appropriate $\lambda $ value is available for perceptual models. In RC, the proposed approach centers on a robust relationship between QP and achieved $\lambda $ . And we also established a bit allocation technique using the related $\lambda $ and expression for the RC model. Experimental results validate the rationality and effectiveness of the proposed method.

Analysis of Video Quality Losses in Homogeneous HEVC Video Transcoding

This paper presents a detailed, quantitative analysis of video quality losses in a homogeneous HEVC video transcoder together with the analysis of the origin of those quality losses and the influence of quantization step alignment on transcoding. With the use of HM15.0 reference software and a set of test video sequences, the cascaded pixel domain video transcoder (CPDT) concept has been used to gather all the necessary data needed for the analysis. This experiment was performed for a wide range of source and target bitrates. The essential result of the work is an extensive evaluation of CPDT, commonly used as a reference in works on effective video transcoding. So far, no such extensively performed study has been made available. The quality degradation between a transcoded video and a video that would be the result of direct compression of the original video at the same bitrate as the transcoded one has been reported. The dependences between quality degradation caused by transcoding and bitrate changes of the transcoded data stream are clearly presented in graphs.

A human vision system-based approach to improve computing efficiency in information systems

While a lot of research has been done in system design, reducing computational complexity is still a core problem in information systems. In order to break through the bottleneck of computing efficiency, this paper draws on the powerful processing capability of the human vision system (HVS) to propose a pixel domain computing framework. It converts the original problem into pixel operations, and pixels are taken as the basic unit to change the construction process of data analysis. The framework significantly decreases computational complexity, simplifies programming logic, and enhances the automaticity of information system. Our results demonstrate that the efficiency of the containment test is improved by 30 times, the logic complexity of the intersection area measurement is reduced by a factor of five, and the automaticity of root finding is significantly heightened.

Deep Image Compression in the Wavelet Transform Domain Based on High Frequency Sub-Band Prediction

In this paper, we propose to use deep neural networks for image compression in the wavelet transform domain. When the input image is transformed from the spatial pixel domain to the wavelet transform domain, one low-frequency sub-band (LF sub-band) and three high-frequency sub-bands (HF sub-bands) are generated. Low-frequency sub-band is firstly used to predict each high-frequency sub-band to eliminate redundancy between the sub-bands, after which the sub-bands are fed into different auto-encoders to do the encoding. In order to further improve the compression efficiency, we use a conditional probability model to estimate the context-dependent prior probability of the encoded codes, which can be used for entropy coding. The entire training process is unsupervised, and the auto-encoders and the conditional probability model are trained jointly. The experimental results show that the proposed approach outperforms JPEG, JPEG2000, BPG, and some mainstream neural network-based image compression. Furthermore, it produces better visual quality with clearer details and textures because more high-frequency coefficients can be reserved, thanks to the high-frequency prediction.

Screen Content Image Quality Assessment With Edge Features in Gradient Domain

Objective visual quality assessment specific for screen content images (SCIs) has been increasingly investigated over the years. In this paper, an effective full-reference quality evaluation model for SCIs is proposed, in which edge features in gradient domain (EFGD) are extracted for better visual perceptual representation. Unlike traditional edge feature extraction directly in the image pixel domain, all edge features in the proposed EFGD model are extracted based on the gradient map of input SCIs, including edge sharpness, edge brightness/contrast, and edge chrominance. Specifically, the gradient profile model that can well represent the spatial layout of edges is adopted to measure the edge sharpness degree. A novel computation way is reported to measure the edge brightness and contrast change between the reference and distorted SCIs, while color moments are used to account for the color chrominance variation. In addition, an adaptive weighting strategy is designed to adjust the effects of these three kinds of edge features, according to the statistical distributions of the input SCIs. Moreover, the maximum value of edge sharpness features is extracted from the test SCIs as the pooling weight to get the final image quality assessment (IQA) score. The experimental results on two commonly used SCIs databases have verified the superiorities of the EFGD model and show that the EFGD model is in more conformity with the subjective assessment results than most of the existing IQA models.

Vehicle-Type Detection Based on Compressed Sensing and Deep Learning in Vehicular Networks.

Throughout the past decade, vehicular networks have attracted a great deal of interest in various fields. The increasing number of vehicles has led to challenges in traffic regulation. Vehicle-type detection is an important research topic that has found various applications in numerous fields. Its main purpose is to extract the different features of vehicles from videos or pictures captured by traffic surveillance so as to identify the types of vehicles, and then provide reference information for traffic monitoring and control. In this paper, we propose a step-forward vehicle-detection and -classification method using a saliency map and the convolutional neural-network (CNN) technique. Specifically, compressed-sensing (CS) theory is applied to generate the saliency map to label the vehicles in an image, and the CNN scheme is then used to classify them. We applied the concept of the saliency map to search the image for target vehicles: this step is based on the use of the saliency map to minimize redundant areas. CS was used to measure the image of interest and obtain its saliency in the measurement domain. Because the data in the measurement domain are much smaller than those in the pixel domain, saliency maps can be generated at a low computation cost and faster speed. Then, based on the saliency map, we identified the target vehicles and classified them into different types using the CNN. The experimental results show that our method is able to speed up the window-calibrating stages of CNN-based image classification. Moreover, our proposed method has better overall performance in vehicle-type detection compared with other methods. It has very broad prospects for practical applications in vehicular networks.

Particle filter-based modulation domain infrared targets tracking

Faced with problems of low contrast, poor SNR, and relatively complicated tracking environment, stable infrared target tracking is worth researching for its many potential applications. In this paper, instead of traditional target tracking in the pixel domain, we propose a sampling importance resampling (SIR) particle filter method with indirect velocity measurements to track infrared targets in the modulation domain. The dominant amplitude modulation (AM) features used for tracking is extracted by decomposing the input image using an 18-channel Gabor filter bank followed by the application of the dominant component analysis approach. The dominant AM modulation features provide a significant partial texture characteristic of the target which can be separated from background with better discrimination. To take advantage of observed kinematics, we utilize the augmented state vector with indirect velocity information via combining the measurements of velocity in adjacent frames to the SIR particle filter framework, which weakens weights of particles with bad velocity estimates but still having association with the cluttered background or other moving objects. A dynamic template update strategy is also provided to prevent the tracker from appearance model drift. Experiments indicate that the proposed method is effective for raising the tracking accuracy compared with other tracking methods.

Pixel‐domain adversarial examples against CNN‐based manipulation detectors

An attack method against convolutional neural network (CNN) detectors, which minimises the distortion in the pixel domain, is proposed. By focusing on CNN models developed for manipulation detection, experiments show that, while the small perturbations introduced by existing methods tend to be cancelled out when the adversarial examples are rounded to pixels, thus making the attack ineffective, the proposed approach can generate pixel-domain adversarial images which succeed in inducing a wrong decision with very small distortions.

Compressed Domain Video Abstraction Based on I-Frame of HEVC Coded Videos

Video abstraction allows indexing, searching, browsing and evaluating a video only by accessing its useful contents. Several studies have been done in this field, but most of them are in pixel domain and require decoding process. It makes these methods more time and process consuming than compressed domain video abstraction. In this paper, we present a new video abstraction method in H.265/HEVC compressed domain, HVAIF. The method is based on the normalized histogram of extracted I-frame prediction modes from an H.265/HEVC coded video. The frames’ similarity is calculated by intersecting their I-frame prediction modes’ histogram. The similarity measure detects and removes redundant key-frames to increase the quality of final video abstraction. Moreover, we employ fuzzy c-means clustering to categorize similar frames and extract key-frames as representatives of the entire video frames. The interpretation of the results shows that using the proposed method achieves on average 86% accuracy and 19% error rate in compressed domain video abstraction which is higher than the other tested methods in the pixel domain. Also, it has an acceptable robustness to coding parameters, and on average it generates video key-frames that are closer to human summaries.

Fast intra coding unit partition decision in H.266/FVC based on spatial features

With the development of technology, the requirements of hardware equipment and user expectations of visual enjoyment are increasingly gradually. The Joint Video Exploration Team (JVET) has established the latest video compression standard, Future Video Coding (FVC). FVC adopts QuadTree plus Binary Tree (QTBT) based Coding Unit (CU) structure, which not only removes the complex hierarchical structure of the CU, Prediction Unit (PU), and Transform Unit (TU) but also supports square and rectangular coding blocks based on the texture of the video content. Although the QTBT structure can provide superior coding performance, it significantly increases the encoding time, particularly in intra coding. Therefore, developing a fast intra CU partition decision algorithm is essential. In this paper, a fast CU partition decision algorithm in FVC intra coding based on spatial features is proposed. Different spatial features in the pixel domain are proposed in the binary tree and quadtree decision processes. Spatial features for the binary tree are employed for early skipping of the encoding process of CUs with binary tree depth and for early determination of binary tree split mode. Spatial features for the quadtree are employed for early splitting or termination of CUs with quadtree depth. Compared with JEM 5.0, the proposed method can save 23% encoding time on average with a slight increase of 0.62% in the Bjontegaard delta bitrate (BDBR).

The Temperature Vegetation Dryness Index (TVDI) Based on Bi-Parabolic NDVI-Ts Space and Gradient-Based Structural Similarity (GSSIM) for Long-Term Drought Assessment Across Shaanxi Province, China (2000–2016)

Traditional NDVI-Ts space is triangular or trapezoidal, but Liu et al. (2015) discovered that the NDVI-Ts space was bi-parabolic when the study area was covered with low biomass vegetation. Moreover, the numerical value of the indicator was considered in most of the study when the drought conditions in the space domain were evaluated. In addition, quantitatively assessing the spatial-temporal changes of the drought was not enough. In this study, first, we used MODIS NDVI and Ts data to reexamine if the NDVI-Ts space with “time” and a single pixel domain is bi-parabolic in the Shaanxi province of China, which is vegetated with low biomass to high biomass. This is compared with the triangular NDVI-Ts space and one of the well-known drought indexes called the temperature-vegetation index (TVX). The results demonstrated that dry and wet edges exhibited a parabolic shape again in scatter plots of Ts and NDVI in the Shaanxi province, which was linear in the triangular NDVI-Ts space. The Temperature Vegetation Dryness Index (TVDIc) was obtained from bi-parabolic NDVI-Ts andTVDIt was obtained from the triangular NDVI-Ts space and TVX were compared with 10-cm depth relative soil moisture. By estimating the 10-cm depth soil moisture, TVDIc was better than TVDIt, which were all apparently better than TVX. Second, combined with MODIS data, the drought conditions of the study area were assessed by TVDIc between 2000 to 2016. Spatially, the drought in the Shaanxi Province between 2000 to 2016 were mainly distributed in the northwest, North Shaanxi, and the North and East Guanzhong plain. The drought area of the Shaanxi province accounted for 31.95% in 2000 and 27.65% in 2016, respectively. Third, we quantitatively evaluated the variation of the drought status by using Gradient-based Structural Similarity (GSSIM) methods. The area of the drought conditions significantly changed and moderately changed at 5.34% and 40.22%, respectively, between 2000 and 2016. Finally, the possible reasons for drought change were discussed. The change of precipitation, temperature, irrigation, destruction or betterment of vegetation, and the enlargement of opening mining, etc., can lead to the variations of drought.

Optimal digital design of steerable differentiators with the flatness of polynomial filters and the isotropy of Gaussian filters

Shape- and scale-selective digital-filters, with steerable finite/infinite impulse responses (FIR/IIRs) and non-recursive/recursive realizations, that are separable in both spatial dimensions and adequately isotropic, are derived. The filters are conveniently designed in the frequency domain via derivative constraints at dc, which guarantees orthogonality and monomial selectivity in the pixel domain (i.e. vanishing moments), unlike more commonly used FIR filters derived from Gaussian functions. A two-stage low-pass/high-pass architecture, for blur/derivative operations, is recommended. Expressions for the coefficients of a low-order IIR blur filter with repeated poles are provided, as a function of scale; discrete Butterworth (IIR), and colored Savitzky-Golay (FIR), blurs are also examined. Parallel software implementations on central processing units (CPUs) and graphics processing units (GPUs), for scale-selective blob-detection in aerial surveillance imagery, are analyzed. It is shown that recursive IIR filters are significantly faster than non-recursive FIR filters when detecting large objects at coarse scales, i.e. using filters with long impulse responses; however, the margin of outperformance decreases as the degree of parallelization increases.

Image deblocking via joint domain learning

Image deblocking is a postprocessing method that aims to suppress the compression artifacts without changing existing JPEG coding standard. We propose an image deblocking method, which is based on deep convolutional neural networks. The proposed method takes full advantage of the characteristics of wavelet domain and pixel domain to restore the high-frequency information of compressed images and maintain low-frequency information, respectively. In addition, a fusion layer is employed to fuse the merits of two domains. Extensive experiments demonstrate that the proposed method outperforms the state-of-the-art deblocking methods in both subjective vision and objective evaluation.

A new secure and robust watermarking technique based on logistic map and modification of DC coefficient

The proliferation of information and communication technology has made exchange of information easier than ever. Security and copyright protection of multimedia contents in such a scenario has become a major challenge for the research community round the globe. Digital watermarking has been found as an effective tool for protection and security of multimedia content. A secure and robust watermarking scheme based on DC coefficient modification in pixel domain and a modified logistic map is presented in this paper. The cover image is divided into $$8 \times 8$$ sub-blocks and instead of computing DC coefficient using discrete cosine transform (DCT), we compute DC coefficient of each block in spatial domain. Watermark bits are embedded by modifying DC coefficients of various blocks in spatial domain. The quantum of change to be brought in various pixels of a block for embedding watermark bit depends upon DC coefficient of respective blocks, nature of watermark bit (0 or 1) to be embedded and the adjustment factor. The security of embedded watermark has been taken care of by using chaotic encryption based on a generalized logistic map (GLM). We show that GLM has better properties like ergodicity, larger lyapunov exponent, uniform invariant density, mixing, higher range of bifurcation parameter etc., compared to basic logistic map. We exploit these properties of GLM for designing a secure robust, strong efficient cryptosystem to encrypt the watermark information before embedding it. Experimental investigations show that besides being highly secure the proposed technique is robust to both signal processing and geometric attacks. Further, the proposed scheme is computationally efficient as DC coefficient which holds the information is computed in pixel domain instead of using DCT on an image block.

An adaptive fuzzy filter for image denoising

This study considers the problem of fuzzy modeling of the images in pixel domain. A zero-order Takagi–Sugeno type fuzzy model provides fuzzy smoothing to the image intensities for removing the additive noise from an image. An adaptive fuzzy filtering algorithm is suggested for estimating the parameters of the fuzzy model with noisy image data. The mathematical analysis of the proposed filtering algorithm has been provided in both deterministic and stochastic framework. The deterministic robustness of the filtering algorithm was shown by deriving an upper bound on the magnitude of estimation errors. The fuzzy filtering algorithm doesn’t demand Gaussian assumption of the noise and is also optimal in the “sense” of variation Bayes towards Student-t distributed noises.

Secure and privacy-preserving data sharing in the cloud based on lossless image coding

Image and video processing in the encrypted domain has recently emerged as a promising research area to tackle privacy-related data processing issues. In particular, reversible data hiding in the encrypted domain has been suggested as a solution to store and manage digital images securely in the cloud while preserving their confidentiality. However, although efficiency has been claimed with reversible data hiding techniques in encrypted images (RDHEI), reported results show that the cloud service provider cannot add more than 1 bit per pixel (bpp) of additional data to manage stored images. This paper highlights the weakness of RDHEI as a suggested approach for secure and privacy-preserving cloud computing. In particular, we propose a new, simple, and efficient approach that offers the same level of data security and confidentiality in the cloud without the process of reversible data hiding. The proposed idea is to compress the image via a lossless image coder in order to create space before encryption. This space is then filled with a randomly generated sequence and combined with an encrypted version of the compressed bit stream to form a full resolution encrypted image in the pixel domain. The cloud service provider uses the created room in the encrypted image to add additional data and produces an encrypted image containing additional data in a similar fashion. Assessed with the lossless Embedded Block Coding with Optimized Truncation (EBCOT) algorithm on natural images, the proposed scheme has been shown to exceed the capacity of 3 bpp of additional data while maintaining data security and confidentiality.