No-Reference Image Quality Assessment with Local Gradient Orientations

Abstract

Image processing methods often introduce distortions, which affect the way an image is subjectively perceived by a human observer. To avoid inconvenient subjective tests in cases in which reference images are not available, it is desirable to develop an automatic no-reference image quality assessment (NR-IQA) technique. In this paper, a novel NR-IQA technique is proposed in which the distributions of local gradient orientations in image regions of different sizes are used to characterize an image. To evaluate the objective quality of an image, its luminance and chrominance channels are processed, as well as their high-order derivatives. Finally, statistics of used perceptual features are mapped to subjective scores by the support vector regression (SVR) technique. The extensive experimental evaluation on six popular IQA benchmark datasets reveals that the proposed technique is highly correlated with subjective scores and outperforms related state-of-the-art hand-crafted and deep learning approaches.Symmetry 2019, 11, 95 2 of 20 patches [28] can be found in the literature.Since the supervised learning bridges image statistics with the perceptual quality, it is often applied to obtain a model used for the quality prediction. For the learning, the Support Vector Regression (SVR) [20,21,23,27], neural networks [24], or random forests [29] are applied. In methods that do not use supervised learning, distortion types are modeled with a set of centroids of quality levels or NSS from multiple cues [30,31]. Another direction is to employ a pseudo-reference image which is created and compared with a distorted image with blockiness, sharpness, and noisiness metrics [32]. Recently, many NR-IQA approaches which use deep neural network (DNN) architectures have been introduced. They merge the feature extraction and quality prediction steps. However, they suffer from a small number of training examples available in IQA benchmark datasets or use complex architectures that are devoted to image recognition tasks. To overcome these limitations, most of them use image patches [33,34], train models using FR-IQA measures instead of subjective scores [34,35], or perform fine-tuning to adopt an architecture to the IQA [36]. Interestingly, some DNN-based approaches use features introduced in earlier methods [35].The HVS is sensitive to local structures, which are often described using local binary patterns (LBP) and gradient-based statistics. However, a spatial distribution of LBP may not be able to capture more complex structures [37]. Thus, statistics extracted from gradient maps often occur in conjunction with other approaches to improve the IQA performance [27,38]. Such techniques use global distributions of gradient magnitude maps [25], relative gradient orientations or magnitude [24,39].To describe an image and efficiently take into account local gradient orientations, Histogram of Oriented Gradients (HOG) descriptor can be used [40]. However, the HOG produces high-dimensional feature vectors, which are devoted to object recognition...

No‐reference image quality assessment of magnetic resonance images with high‐boost filtering and local features

“…The method was compared with the following 21 state‐of‐the‐art NR techniques with publicly available sourcecode: BRISQUE, NOREQI, BPRI, IL‐NIQE, HOSA, GWHGLBP, ORACLE, RATER, SCORER, QENI, GM‐LOG, SISBLIM, metricQ, SSEQ, S‐INDEX, NFERM, SEER, DEEPIQ, MEON, WaDIQaM‐NR, and SNRTOI . The NOREQI, BPRI, ORACLE, RATER, QENI, and SCORER are the most similar to NOMRIQA since they employ local features.…”

“…These measures represent a variety of approaches which have emerged in the IQA literature, including free‐energy principle, entropy, MSCN coefficients, deep learning methods, or distortion‐specific techniques . They also include previous authors’ works on IQA of natural images which employ gradient operators with real‐valued local features and image statistics, optimization‐based filters, derivative‐based filters with a quality‐aware feature descriptor, high‐pass filters and statistics of obtained binary descriptors, statistics of global descriptors considering differently defined local neighborhoods, or combined visual saliency models of image patches with self‐similarity of real‐valued local descriptors …”

Fusion of Deep Convolutional Neural Networks for No-Reference Magnetic Resonance Image Quality Assessment

“…As gradient-based features can effectively describe distorted images, many approaches use them for quality prediction. They employ global distributions of gradient magnitude maps [18], relative gradient orientations or magnitude [19], and local gradient orientations captured by Histogram of Oriented Gradient (HOG) technique for variously defined neighborhoods [20]. A histogram of local binary patterns (LBP) characterizing a gradient map of an image is used in the GWHGLBP approach [21].…”

“…The introduced approach is represented by three fusion models: Resnet-50_GoogLeNet _ResNet-18 (R50GR18), ResNet-50_GoogLeNet_MobileNet-V2 (R18GR50M), and MobileNet-V2_ResNet-50 (MR50). They are experimentally compared with 17 state-of-the-art techniques: NFERM [47], SEER [20], DEEPIQ [27], MEON [46], SNRTOI [48], NOREQI [22], BPRI [24], HOSA [17], NOMRIQA [8], IL-NIQE [23], GM-LOG [18], GWHGLBP [21], BRISQUE [15], SISBLIM [49], metricQ [50], SINDEX [51], and ENMIQA [7]. The NOM-RIQA, ENMIQA, and SNRTOI are designed for MR images, whereas DEEPIQ and MEON are deep learning approaches devoted to natural images.…”

Current Trends and Advances in Image Quality Assessment

“…One of the most recent ideas is the application of the analysis of distributions of local gradient orientations in image regions of different sizes [69], outperforming even some deep learning based IQA methods.…”