NATURAL SCENE STATISTICS AND CNN BASED PARALLEL NETWORK FOR IMAGE QUALITY ASSESSMENT

Image Quality Assessment (IQA) tasks have increasing importance in today's world due to the widespread use of imaging devices and social media. Statistical studies proved that naturalness measures are good discriminators for evaluating image quality. Convolutional neural networks (CNN) based IQA models gained popularity in recent years due to their enhanced performance. In this article, we present a no-reference image quality assessment method that integrates natural image statistics (NSS) with CNN. The proposed approach extracts NSS features from the image, integrates them with the CNN features to predict the quality score. Our experimental results show that the performance of the proposed method is competitive against the existing methods of image quality assessment. Cross database testing on Live in the Wild (LIVE-itW) and Smartphone Photography Attribute and Quality (SPAQ) databases shows excellent generalization.