Image Quality Assessment using Selective Contourlet Coefficients
Agnel Lazar Alappat, Vipin Milind Kamble · 2020
The ubiquity of digital images getting captured and viewed everyday in today's world continuously increases the importance of image processing or image quality assessment to be precise. In this paper, we present an algorithm for Non-Referral Image Quality Assessment (NR-IQA). We make use of the information available at various levels of contourlet transform of a degraded image to predict the level of distortion. We use the simple statistical values of the contourlet coefficients of images to train Neural Network Model which gives us the level of degradation present in the corresponding image. This variable extraction and training algorithm has been found to show high correlation with standard datasets. Benchmarking and validation on the TID2013 dataset indicates that the proposed algorithm outperforms the state of the art algorithms.