No-reference image quality assessment using statistical characterization in the shearlet domain

作者：

Highlights：

• Natural scene statistics (NSS) model in the shearlet domain has been proposed to blindly measure the quality of images.

• A robust prediction algorithm is also proposed based on NSS model, which is suitable for many common distortions.

• An algorithm is proposed to quantify and calibrate image quality by translating NR-IQA into classification problem.

摘要

•Natural scene statistics (NSS) model in the shearlet domain has been proposed to blindly measure the quality of images.•A robust prediction algorithm is also proposed based on NSS model, which is suitable for many common distortions.•An algorithm is proposed to quantify and calibrate image quality by translating NR-IQA into classification problem.

论文关键词：No-reference image quality assessment,Shearlet transform,Sparse autoencoder,Softmax classification

论文评审过程：Received 4 January 2014, Revised 26 April 2014, Accepted 28 May 2014, Available online 11 June 2014.

论文官网地址：https://doi.org/10.1016/j.image.2014.05.007