Blind Image Quality Assessment Using Convolutional Neural Network

Abstract

In this paper, we use Convolutional Neural Network for Blind Image Quality Assessment (BIQA) by utilizing its power to extract features from images and then learn a score or quality index for each image. The evaluation of the proposed model conducted on TID2013 database reveals that using CNN model is way more effective in assessing the quality of images with various distortions in comparison to the other existing assessment methods. The Spearman Rank-Order Correlation Coefficient, used to evaluate the performance of the model, has a very high value in comparison to other existing models, suggesting the efficiency of the proposed model.