Figure 10 - uploaded by Andrew B. Watson
Content may be subject to copyright.
JND errors pooled over frames and blocks. Each panel shows the pooled error at each DCT frequency for each color, over the complete sequence.
Source publication
The growth of digital video has given rise to a need for computational methods for evaluating the visual quality of digital video. We have developed a new digital video quality metric, which we call DVQ (digital video quality) [A. B. Watson, in Human Vision, Visual Processing, and Digital Display VIII, Proc. SPIE3299, 139–147 (1998)]. Here, we prov...
Context in source publication
Citations
... In human vision, the area of interest is usually focused on the boundary changes of an image [46]. Polarization calculation imaging can provide rich contour texture information, especially in the vicinity of the object, as shown in Fig. 7 to Fig. 13. ...
Traditional optical imaging relies on light intensity information from light reflected or transmitted by an object, while polarization imaging utilizes polarization information of light. Camera array imaging is a potent computational imaging technique that enables computational imaging at any depth. However, conventional imaging methods mainly focus on removing occlusions in the foreground and targeting, with limited attention to imaging and analyzing polarization characteristics at specific depths. Conventional camera arrays cannot be used for polarization layered computational imaging. Thus, to study polarization layered imaging at various depths, we devised a flexible polarization camera array system and proposed a depth-parallax relationship model to achieve computational imaging of polarization arrays and polarization information reconstruction under varying conditions and depths. A series of experiments were conducted under diverse occlusion environments. We analyzed the distinctive characteristics of the imaging results obtained from the polarization array, employing a range of array distribution methods, materials, occlusion density, and depths. Our research successfully achieved computational imaging that incorporates a layered perception of objects. Finally, we evaluated the object region’s polarization information using the gray level co-occurrence matrix feature method.
... 1], some of which are based on the properties of human vision [e.g. 2,3] (see [4] for a review). However, the general view is that the "gold standard" is a human observer [5] which means that human judgements are required. ...
Image processing algorithms are used to improve digital image representations in either their appearance or storage efficiency. The merit of these algorithms depends, in part, on visual perception by human observers. However, in practice, most are assessed numerically, and the perceptual metrics that do exist are criterion sensitive with several shortcomings. Here we propose an objective performance-based perceptual measure of image quality and demonstrate this by comparing the efficacy of a denoising algorithm for a variety of filters. For baseline, we measured detection thresholds for a white noise signal added to one of a pair of natural images in a two-alternative forced-choice (2AFC) paradigm where each image was selected randomly from a set of n = 308 on each trial. In a series of experimental conditions, the stimulus image pairs were passed through various configurations of a denoising algorithm. The differences in noise detection thresholds with and without denoising are objective perceptual measures of the ability of the algorithm to render noise invisible. This was a factor of two (6dB) in our experiment and consistent across a range of filter bandwidths and types. We also found that thresholds in all conditions converged on a common value of PSNR, offering support for this metric. We discuss how the 2AFC approach might be used for other algorithms including compression, deblurring and edge-detection. Finally, we provide a derivation for our Cartesian-separable log-Gabor filters, with polar parameters. For the biological vision community this has some advantages over the more typical (i) polar-separable variety and (ii) Cartesian-separable variety with Cartesian parameters.
... Because of these factors, it is critical that students begin to develop essential math concepts and skills at an early age. From the practical perspective, mathematical tasks are the bedrock of mathematics lessons in many countries (Watson & Ohtani, 2015). Mathematics classroom instruction is normally organized and delivered through the activities in mathematical tasks developed by teachers or found in curriculum materials. ...
... To examine classroom instructions, it is common to look at and analyze the amount of lesson time that students spend on doing the task (Shimizu, Kaur, Huang, & Clarke, 2010). Watson and Ohtani (2015) also point out, from the cognitive viewpoint, that the goal and content of a mathematical task are important and can have significant effect on students' learning. ...
... Some image quality metrics try to mimic the human perception and either predict the perceived image quality or perceived similarity between two images. his category includes such metrics as SSIM [48], ISSIM [49], DVQ [50] and others. Each of these metrics estimate in one way or another the expressiveness of object features in the image. ...
Computed tomography is an important technique for non-destructive analysis of an object’s internal structure, relevant for scientific studies, medical applications, and industry. Pressing challenges emerging in the field of tomographic imaging include speeding up reconstruction, reducing the time required to obtain the X-ray projections, and reducing the radiation dose imparted to the object. In this paper, we introduce a model of a monitored reconstruction process, in which the acquiring of projections is interspersed with image reconstruction. This model allows to examine the tomographic reconstruction process as an anytime algorithm and consider a problem of finding the optimal stopping point, corresponding to the required number of X-ray projections for the currently scanned object. We outline the theoretical framework for the monitored reconstruction, propose ways of constructing stopping rules for various reconstruction quality metrics and provide their experimental evaluation. Due to stopping at different times for different objects, the proposed approach allows to achieve a higher mean reconstruction quality for a given mean number of X-ray projections. Conversely, fewer projections on average are used to achieve the same mean reconstruction quality.
... Some image quality metrics try to mimic the human perception and either predict the perceived image quality or perceived similarity between two images. his category includes such metrics as SSIM [48], ISSIM [49], DVQ [50] and others. Each of these metrics estimate in one way or another the expressiveness of object features in the image. ...
The purpose of this research is to develop a novel approach to benchmark smart community health centers in order to achieve continuous improvement of service quality. Three methods are presented: the fuzzy DEMATEL method is used to determine the criteria weights, the fuzzy ELECTRE III method is employed to obtain the ranking of smart community health centers, and IPA (Importance-Performance Analysis) is employed to formulate improvement strategies. The proposed approach clearly identifies the strengths and weakness of each smart community health center by ranking its performance with respect to a system of five service quality criteria. In addition, IPA is able to develop the most effective improvement strategies for each smart community health center. The proposed approach was applied to five smart community health centers in Beijing and service strengths and weakness are discussed. The proposed approach has three notable advantages. First, the novel approach can address ambiguity and uncertainty in the process of decision making. Second, interdependent relationships among the evaluation criteria are analyzed by the fuzzy DEMATEL method, so that the weights obtained are more in line with reality. Third, the fuzzy ELECTRE III method considers non-compensatory behavior for service quality comparisons among smart community health centers. The novel fuzzy-based approach presented in this paper is a powerful and highly effective tool to benchmark smart community health centers and develop successful improvement strategies of service quality.
... , VIF27 , MAD28 , SSIM 29 , FSIM 30 , NLOG 31 , VSI 32 , PAMSE33 , and NSER 34 , can perform well for the spatial feature extraction of video. For the temporal feature extraction, most QA algorithms usually use simple average pooling image sequence, frame difference features, and temporal features from a group of frames[35][36][37] . ...
... This is followed by the extraction of perception based features [44], computation of video quality parameters, and calculation of VQM models. Watson et al. had proposed Digital Video Quality (DVQ) [43] which embodies just noticeable difference (JND). In DVQ, discrete cosine transform (DCT) coefficients of a video are first filtered by contrast sensitivity function (CSF). ...
Videos are amongst the most popular online media for Internet users nowadays. Thus, it is of utmost importance that the videos transmitted through the internet or other transmission media to have a minimal data loss and acceptable visual quality. Video quality assessment (VQA) is a useful tool to determine the quality of a video without human intervention. A new VQA method, termed as Error and Temporal Structural Similarity (EaTSS), is proposed in this paper. EaTSS is based on a combination of error signals, weighted Structural Similarity Index (SSIM) and difference of temporal information. The error signals are used to weight the computed SSIM map and subsequently to compute the quality score. This is a better alternative to the usual SSIM index, in which the quality score is computed as the average of the SSIM map. For the temporal information, the second-order time-differential information are used for quality score computation. From the experiments, EaTSS is found to have competitive performance and faster computational speed compared to other existing VQA algorithms.
... Both of them are popular because of their high efficiency and ease of use, however, their inconsistency with subjective perception is often criticized. Therefore, researchers have tried to study algorithms that are consistent with subjective perception and proposed various objective algorithms, such as the FR VQA algorithms Content-weighted VQA (3-SSIM) [1] , temporal quality variations (TQV) [2] , digital video quality metric(DVQ) [3] , perceptual quality index (PQI) [4] , temporal trajectory aware video quality (Tetra VQM) [5] , video quality metric (VQM) [6] , Video structure similarity measure (VSSIM) [7] , spatiotemporal most-apparent-distortion measure (ST-MAD) [8] , MOTION-based video integrity assessment (MOVIE) [9] , VQA analysis via spatial and spatiotemporal slices (ViS3) [10] , VQA via gradient magnitude similarity of spatial and spatiotemporal slices (STS-GMSD/SSTS-GMSD) [11] , motion structure partition similarity (MSPS) [12] , and the NR algorithm blind natural video quality measure (V-BLIINDS) [13] , etc. Some of which will be briefly reviewed below. ...
... For the spatial feature extraction of videos, classic IQA algorithms such as GMSD [26] , VIF [27] , MAD [14] , SSIM [28] , FSIM [29] , NLOG [30] , VSI [31] , PAMSE [32] , NSER [33] , NLOG [30] etc. have performed well. For the temporal feature extraction and computing, most quality evaluation algorithms often use simple average pooling of image quality, frame difference features, and features come from a small group of frames [1][2][3], [6] . Compared with the previous temporal feature extraction methods, STS images provide the convenience of extracting perceptual features from a larger time scale, which will facilitate the design of effective quality evaluation algorithms, and the success of FR MSPS [12] and ViS3 [10] have illustrated this point. ...
... The work in [8] applied the JND threshold on the DCT coefficients of differences between original and distorted frames. Another method, Digital Video Quality (DVQ) [36], applies the JND threshold on the second order infinite impulse response (IIR) filtered coefficients. Instead of applying JND on the transformed coefficients, JVQ applies the JND threshold directly to the frame differences. ...
A Just Noticeable Difference (JND)-based video quality assessment (VQA) method is proposed. This method, termed as JVQ, applies JND concept to structural similarity (SSIM) index to measure the spatial quality. JVQ incorporates three features, i.e. luminance adaptation, contrast masking, and texture masking. In JVQ, the concept of JND is refined and more features are considered. For the spatial part, minor distortions in the distorted frames are ignored and considered imperceptible. For the temporal part, SSIM index is simplified and used to measure the temporal video quality. Then, a similar JND concept which comprises of temporal masking is also applied in the temporal quality evaluation. Pixels with large variation over time are considered as not distorted because the distortions in these pixels are hardly perceivable. The final JVQ index is the arithmetic mean of both spatial and temporal quality indices. JVQ is found to achieve good correlation with subjective scores. In addition, this method has low computational cost as compared to existing state-of-the-art metrics.
... A second group of FR and RR objective quality metrics are those specifically designed for video quality assessment, taking into account aspects of both spatial and temporal domains. The DCT-VQM (Discrete Cosine Transform-Video Quality Metric) metric [26] (based on Watson's proposal in [27]) operates in the frequency domain of the sequences through a DCT. It takes into account the decrease in human visual sensitivity at high spatial and temporal frequencies. ...
Quality of experience is of critical importance in streaming video services, because the traditional quality of service cannot represent the quality perceived by viewers. This work evaluates several objective quality metrics under realistic bursty packet loss conditions in the network, with the support of a packet loss model. Alignment of reference and streamed video sequences (with different levels of spatial-temporal information) are also explored as a technique to prevent inaccurate computation of objective metrics due to frame loss. Finally, the correlation between subjective and objective metrics for each motion level and the computing time of metrics are analysed. The most suitable objective metrics to characterize the real degradation in the quality perceived by viewers, for both off-line and real-time assessment, are proposed. The integration of motion, busty packet loss, sequence alignment after frame loss and computing time of metrics are the main contributions of this research work.
