Blind Image Quality Assessment: From Natural Scene Statistics to Perceptual Quality

By Anush K. Moorthy, A. Bovik. 2011

1,743 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Media Technology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

The Perception-Distortion Tradeoff (2017, methodology): “Perceptual quality, on the other hand, is ideally quantified by collecting human opinion scores, which is time consuming and costly [27, 33].”
Learning a No-Reference Quality Metric for Single-Image Super-Resolution (2016, methodology): “For DIVINE and BIQI, we present the reported results as the performance of these methods by retraining on our dataset is significantly worse.”
TOPIQ: A Top-Down Approach From Semantics to Distortions for Image Quality Assessment (2023, methodology): “Traditional approaches for NR-IQA rely on natural scene statistics (NSS) [3], [26]–[30].”
RankIQA: Learning from Rankings for No-Reference Image Quality Assessment (2017, background): “N R -I Q A DIVINE [21] 0.”
Convolutional Neural Networks for No-Reference Image Quality Assessment (2014, methodology): “Traditional NSS based features are extracted in image transformation domains using, for example the wavelet transform [10] or the DCT transform [13].”
A Completely Blind Video Integrity Oracle (2016, methodology): “significantly improve predictions of the perceptual impact of statistical impairments as can be seen, for example by the favorable performance of the spatially isotropic model [31] relative to the orientation-selective DIIVINE index [36].”
No-Reference Image Quality Assessment via Transformers, Relative Ranking, and Self-Consistency (2021, background): “Variations of NSS features in different domains such as spatial [6, 18, 8], gradient [8], discrete cosine transform (DCT) [5], and wavelet [17], showed impressive performances for synthetically distorted images.”
KonIQ-10k: An Ecologically Valid Database for Deep Learning of Blind Image Quality Assessment (2019, methodology): “These methods were BIQI [32], BLIINDS-II [35], BRISQUE [36], CORNIA [7], DIIVINE [33], HOSA [8], and SSEQ [34].”
Reversible data hiding: Advances in the past two decades (2016, background): “, [149]–[154]) for image quality assessment.”
No-Reference Image Quality Assessment in the Spatial Domain (2012, methodology): “SVR has previously been applied to image quality assessment problems [12], [37], [38].”
dipIQ: Blind Image Quality Assessment by Learning-to-Rank Discriminable Image Pairs (2017, methodology): “Therefore, statistics of raw [4], [6], [46], [47] and normalized [48], [49] wavelet coefficients, and wavelet coefficient correlations in the neighborhood [10], [29], [50]–[52] can be individually or jointly modeled as image naturalness measurements.”
Using Free Energy Principle For Blind Image Quality Assessment (2015, methodology): “We first evaluate the performance of the proposed NFERM with a total number of fifteen IQA approaches on the LIVE database [39]: 1) classical FR PSNR, SSIM [4], MS-SSIM [5]; 2) popular blind distortion-specific WNJE [21], MBBM [22], Sheikh [23], SINE [24], JNB [26], CPBD [27], NFEQM [17]; 3) state-o”
MUSIQ: Multi-scale Image Quality Transformer (2021, background): “But since each crop is only method SRCC PLCC DIIVINE [28] 0.599 0.600 BRISQUE [26] 0.809 0.817 CORNIA [42] 0.709 0.725 QAC [40] 0.092 0.497 ILNIQE [47] 0.713 0.721 FRIQUEE [14] 0.819 0.830 DBCNN [49] 0.911 0.915 Fang et al .”
Deep CNN-Based Blind Image Quality Predictor (2019, methodology): “All the subjective scores were rescaled to [0, 1], and DMOSs were reversed to match with MOSs.”
Blind/Referenceless Image Spatial Quality Evaluator (2011, methodology): “Two recently developed algorithms BLind Image Notator using DCT Statistics (BLIINDS) - II [12] and the Distortion Identification-based Image Verity and INtegrity Evaluation (DIIVINE) were developed along these lines, making no assumption on the kind of distortion medium image is passed through.”
Domain Fingerprints for No-Reference Image Quality Assessment (2019, background): “The existing studies on NR-IQA can be broadly classiﬁed into two categories: designing hand-crafted features [1], [2], [10], [11] and learning discriminant visual features automatically [7]–[9].”
Group MAD Competition? A New Methodology to Compare Objective Image Quality Models (2016, methodology): “Fifth, a model that is worth noting is LPSI [36], which essentially reduces the feature space to one dimension and without using MOS for training, but it outperforms sophisticated machine learning-based approaches such as BRISQUE [17] and DIIVINE [20] which use many features for training.”
Perceptual quality prediction on authentically distorted images using a bag of features approach (2016, methodology): “…machine learning methods: Although using NSS models for IQA remains an active research area (Mittal, Moorthy, & Bovik, 2012; Zhang et al., 2014; Moorthy & Bovik, 2011; Goodall, Bovik, & Paulter, 2016; Zhang & Chandler, 2013), our bag of features approach goes signiﬁcantly beyond through the…”