An Image Quality Index Based on Mutual Information and Neural Networks

Measurement of image and video quality is of fundamental importance in a broad range of multimedia applications. The ultimate goal of quality evaluation algorithms is to assess automatically the quality of images or videos in agreement with subjective human judgments. We discuss in this paper, a new approach for measuring image quality across different types of degradations that affect a given image or a video sequence. We start by ranking different image quality indices, traditionally used, based on their information content. Then, we introduce a neural network approach based on the top-ranked indices to predict the Mean Opinion Score of human observers. The experimental results show that the proposed composite quality index results in superior performance compared to traditional measures when used individually.