Performance Analysis of No Reference Image quality based on Human Perception

Abstract

In this work, a No-Reference objective image quality assessment based on NRDPF-IQA metric and classification based metric are tested using LIVE database, which consisting of Gaussian white noise, Gaussian blur, Rayleigh fast fading channel, JPEG compressed images, JPEG2000 images. We plot the Spearman’s Rank Order Correlation Coefficient [SROCC] between each of these features and human DMOS from the LIVE-IQA database using our proposed method to ascertain how well the features correlate with human judgement quality. The analysis of the testing and training is done by SVM model. The proposed method shows better results compared with the earlier methods. Finally, the results are generated by using MATLAB.DOI:http://dx.doi.org/10.11591/ijece.v4i6.6783