Blind Image Quality Assessment Using Semi-supervised Rectifier Networks

Tang, Huixuan; Joshi, Neel; Kapoor, Ashish

doi:10.1109/cvpr.2014.368

Cited by 91 publications

(58 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Some papers construct the models of human visual system (HVS), which can evaluate the quality of the stereoscopic image [43]. For example, the work [22,24] can evaluate the quality of the stereoscopic image by the combination of the absolute difference of the left and right views on the stereoscopic perception; The other literatures use error calculation (PSNR) as a index to evaluate the quality of the stereoscopic image by simulate multi-channel decomposition characteristics [46,47]. However, At present, the human understanding of HVS is very limited, so some literature use a simple square error algorithm mean, such as the use of the literature in [37] using PSNR method to define index of Stereo stereoscopic quality objective evaluation from image quality and sense of three-dimensional.…”

Section: Introductionmentioning

confidence: 99%

RETRACTED ARTICLE: Quality assessment for virtual reality technology based on real scene

Jiang

Yang

Jiang³

et al. 2016

Neural Comput & Applic

View full text Add to dashboard Cite

Virtual reality technology is a new display technology, which provides users with real viewing experience. As known, most of the virtual reality display through stereoscopic images. However, image quality will be influenced by the collection, storage and transmission process . If the stereoscopic image quality in the virtual reality technology is seriously damaged, the user will feel uncomfortable, and this can even cause healthy problems.In this paper, we establish a set of accurate and effective evaluation for the virtual reality. In the preprocessing, we segment the original reference and distorted image into binocular regions and monocular regions. Then, the IW-SSIM(Information weighted-SSIM) or IW-PSNR(Information weighted-PSNR) values over the monocular regions are applied to obtain the IW-score.At the same time, the SW-SSIM(Stereo weighted-SSIM)or SW-PSNR(Stereo weighted-PSNR) can be used to caculate the SW-score. Finally, we pool the stereoscopic images score by combing the IW-score and SW-score. Experiments show that our method is very consistent with human subjective judgment standard in evaluation of virtual reality technology.

show abstract

Section: Introductionmentioning

confidence: 99%

RETRACTED ARTICLE: Quality assessment for virtual reality technology based on real scene

Jiang

Yang

Jiang³

et al. 2016

Neural Comput & Applic

View full text Add to dashboard Cite

show abstract

“…IQA algorithms could be classified into three categories: full-reference IQA (FR-IQA) [50,24,19], reducedreference IQA (RR-IQA) [11], and general purpose noreference IQA (NR-IQA) [46,17,42,51,44,20,25]. Al- the Ground-truth Reference image which is undistorted.…”

Section: Introductionmentioning

confidence: 99%

Hallucinated-IQA: No-Reference Image Quality Assessment via Adversarial Learning

Lin

Wang²

2018

2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition

207

104

View full text Add to dashboard Cite

No-reference image quality assessment (NR-IQA) is a fundamental yet challenging task in low-level computer vision community. The difficulty is particularly pronounced for the limited information, for which the corresponding reference for comparison is typically absent. Although various feature extraction mechanisms have been leveraged from natural scene statistics to deep neural networks in previous methods, the performance bottleneck still exists.In this work, we propose a hallucination-guided quality regression network to address the issue. We firstly generate a hallucinated reference constrained on the distorted image, to compensate the absence of the true reference. Then, we pair the information of hallucinated reference with the distorted image, and forward them to the regressor to learn the perceptual discrepancy with the guidance of an implicit ranking relationship within the generator, and therefore produce the precise quality prediction. To demonstrate the effectiveness of our approach, comprehensive experiments are evaluated on four popular image quality assessment benchmarks. Our method significantly outperforms all the previous state-of-the-art methods by large margins. The code and model are publicly available on the project page https://kwanyeelin.github. io/projects/HIQA/HIQA.html.

show abstract

“…Recent studies develop the noreference sharpness metric [14,2] with the guidance of subjective image quality database and even unlabeled images. The subjective databases provide the MOS (subjective mean opinion scores) as the ground truth of blur extent, and the unlabeled data are synthesized with the same distortion type and level as the subjective databases [29,27]. However, most databases use the synthetic images, which are generated from a limited set of source images with spatially-invariant Gaussian convolution.…”

Section: Introductionmentioning

confidence: 99%