A Machine Learning based Reduced-Reference Image Quality Assessment Method for Single-Image Super-Resolution

The existing image quality assessment methods can be classified into full-reference, no-reference, and reduced-reference methods. However, full-reference methods require ground-truth images which are unavailable for single-image super-resolution (SISR). No-reference methods do not require reference images, but establishing a model is more complicated. Low-resolution (LR) images are known for SISR, and partial information can be provided as reference. In this paper, we propose a machine learning based reduced-reference image assessment method for SISR. LR images are used as reference images. First, we extract the statistical features of LR images and super-resolution (SR) images in spatial domain and log-Gabor filter response respectively. Then, learn a two-stage regression model to measure the quality of SR images referring to LR images. Compared with the state-of-art methods, experimental results proved that the proposed method is more consistent with human visual perception.

[1]  Lei Zhang,et al.  Blind Image Quality Assessment Using Joint Statistics of Gradient Magnitude and Laplacian Features , 2014, IEEE Transactions on Image Processing.

[2]  Yannick Berthoumieu,et al.  Multiscale skewed heavy tailed model for texture analysis , 2009, 2009 16th IEEE International Conference on Image Processing (ICIP).

[3]  Chih-Yuan Yang,et al.  Learning a No-Reference Quality Metric for Single-Image Super-Resolution , 2016, Comput. Vis. Image Underst..

[4]  Bo Yan,et al.  A Deep Learning Based No-Reference Image Quality Assessment Model for Single-Image Super-Resolution , 2018, 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[5]  Michal Irani,et al.  Super-resolution from a single image , 2009, 2009 IEEE 12th International Conference on Computer Vision.

[6]  Luc Van Gool,et al.  Anchored Neighborhood Regression for Fast Example-Based Super-Resolution , 2013, 2013 IEEE International Conference on Computer Vision.

[7]  Alan C. Bovik,et al.  Making a “Completely Blind” Image Quality Analyzer , 2013, IEEE Signal Processing Letters.

[8]  Lei Zhang,et al.  Image Deblurring and Super-Resolution by Adaptive Sparse Domain Selection and Adaptive Regularization , 2010, IEEE Transactions on Image Processing.

[9]  Arnold W. M. Smeulders,et al.  c ○ 2005 Springer Science + Business Media, Inc. Manufactured in The Netherlands. A Six-Stimulus Theory for Stochastic Texture , 2002 .

[10]  Lei Zhang,et al.  A Feature-Enriched Completely Blind Image Quality Evaluator , 2015, IEEE Transactions on Image Processing.

[11]  Zhe L. Lin,et al.  Fast Image Super-Resolution Based on In-Place Example Regression , 2013, 2013 IEEE Conference on Computer Vision and Pattern Recognition.

[12]  Qian Hu,et al.  Reduced-Reference Image Quality Assessment for Single-Image Super-Resolution Based on Wavelet Domain , 2019, 2019 Chinese Control And Decision Conference (CCDC).

[13]  Xiaoou Tang,et al.  Learning a Deep Convolutional Network for Image Super-Resolution , 2014, ECCV.

[14]  D J Field,et al.  Relations between the statistics of natural images and the response properties of cortical cells. , 1987, Journal of the Optical Society of America. A, Optics and image science.

[15]  Thomas S. Huang,et al.  Image Super-Resolution Via Sparse Representation , 2010, IEEE Transactions on Image Processing.

[16]  Eero P. Simoncelli,et al.  Image quality assessment: from error visibility to structural similarity , 2004, IEEE Transactions on Image Processing.

[17]  Wenhan Yang,et al.  Blind visual quality assessment for image super-resolution by convolutional neural network , 2018, Multimedia Tools and Applications.