MS-UNIQUE: Multi-model and Sharpness-weighted Unsupervised Image Quality Estimation

Prabhushankar, Mohit; Temel, Dogancan; AlRegib, Ghassan

doi:10.2352/issn.2470-1173.2017.12.iqsp-223

Cited by 26 publications

(21 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…x > 0 is a Gamma function. We estimate the parameters α i and σ 2 i from the GGD fit of the three NFMs, and estimate the parameters γ i , σ 2 i,r , andσ 2 i,l from the AGGD fit of pairwise products of neighboring pixels in the three NFMs along four orientations.…”

Section: Distribution Statistics Within Different Receptive Fieldsmentioning

confidence: 99%

“…As the crucial aspect in optimization problems of image processing applications, the image quality assessment (IQA) algorithms aim to automatically and accurately evaluate the quality of a given image without accessing the ground truth [1][2][3][4][5]. Compared with full reference (FR) [6,7] IQA and reduced reference (RR) IQA [8] algorithms, blind IQA (BIQA) algorithms can estimate the perceptual quality of a distorted image without using any information of its pristine image.…”

Section: Introductionmentioning

confidence: 99%

“…3D scatter plots between AGGD parameters α, σ 2 l and σ2 r in horizontal orientation of the three NFMs, where each point in the scatter plots represent one distorted image in LIVE[28] database: (a) scatter plots for the first NFM; (b) scatter plots for the second NFM; and (c) scatter plots for the third NFM.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Blind Image Quality Assessment Based on Multi-Window Method and HSV Color Space

Tang

Jiang

et al. 2019

Applied Sciences

View full text Add to dashboard Cite

To improve the evaluation accuracy of the distorted images with various distortion types, an effective blind image quality assessment (BIQA) algorithm based on the multi-window method and the HSV color space is proposed in this paper. We generate multiple normalized feature maps (NFMs) by using the multi-window method to better characterize image degradation from the receptive fields of different sizes. Specifically, the distribution statistics are first extracted from the multiple NFMs. Then, Pearson linear correlation coefficients between spatially adjacent pixels in the NFMs are utilized to quantify the structural changes of the distorted images. Weibull model is utilized to capture distribution statistics of the differential feature maps between the NFMs to more precisely describe the presence of the distortions. Moreover, the entropy and gradient statistics extracted from the HSV color space are employed as a complement to the gray-scale features. Finally, a support vector regressor is adopted to map the perceptual feature vector to image quality score. Experimental results on five benchmark databases demonstrate that the proposed algorithm achieves higher prediction accuracy and robustness against diverse synthetically and authentically distorted images than the state-of-the-art algorithms while maintaining low computational cost.

show abstract

Section: Distribution Statistics Within Different Receptive Fieldsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Blind Image Quality Assessment Based on Multi-Window Method and HSV Color Space

Tang

Jiang

et al. 2019

Applied Sciences

View full text Add to dashboard Cite

show abstract

“…The PSNR and SSIM are full reference IQA (FR IQA) methods that do not depend upon the data. Recently, data-driven FR IQA methods based on unsupervised learning have been proposed [2], [3], which perform better than the PSNR and SSIM. However, these FR IQA methods require the original images and the distorted image to evaluate the image quality.…”

Section: Introductionmentioning

confidence: 99%

Discriminative Convolutional Neural Network for Image Quality Assessment with Fixed Convolution Filters

Takagi

Sakurai

Hagiwara

2019

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

Current image quality assessment (IQA) methods require the original images for evaluation. However, recently, IQA methods that use machine learning have been proposed. These methods learn the relationship between the distorted image and the image quality automatically. In this paper, we propose an IQA method based on deep learning that does not require a reference image. We show that a convolutional neural network with distortion prediction and fixed filters improves the IQA accuracy.

show abstract

“…Interestingly, in the preprocessing step, a color space selection is performed (conversion into YCbCr model is suggested with replacement of the Cb chrominance by the green channel) followed by random patch sampling, forming the vector containing 64 elements for each of three channels, further normalization using a mean subtraction and a whitening operation. The additional extension by analyzing the learned weights was proposed as the MS-UNIQUE metric [32]. Both metrics were trained using randomly selected patches from the ImageNet database.…”

mentioning

confidence: 99%

Combined Full-Reference Image Quality Metrics for Objective Assessment of Multiply Distorted Images

Okarma

Lech

2021

Electronics

View full text Add to dashboard Cite

In the recent years, many objective image quality assessment methods have been proposed by different researchers, leading to a significant increase in their correlation with subjective quality evaluations. Although many recently proposed image quality assessment methods, particularly full-reference metrics, are in some cases highly correlated with the perception of individual distortions, there is still a need for their verification and adjustment for the case when images are affected by multiple distortions. Since one of the possible approaches is the application of combined metrics, their analysis and optimization are discussed in this paper. Two approaches to metrics’ combination have been analyzed that are based on the weighted product and the proposed weighted sum with additional exponential weights. The validation of the proposed approach, carried out using four currently available image datasets, containing multiply distorted images together with the gathered subjective quality scores, indicates a meaningful increase of correlations of the optimized combined metrics with subjective opinions for all datasets.

show abstract

MS-UNIQUE: Multi-model and Sharpness-weighted Unsupervised Image Quality Estimation

Cited by 26 publications

References 17 publications

Blind Image Quality Assessment Based on Multi-Window Method and HSV Color Space

Blind Image Quality Assessment Based on Multi-Window Method and HSV Color Space

Discriminative Convolutional Neural Network for Image Quality Assessment with Fixed Convolution Filters

Combined Full-Reference Image Quality Metrics for Objective Assessment of Multiply Distorted Images

Contact Info

Product

Resources

About