ABSTRACT
In recent years, the need to accurate and reliable objective Image Quality Assessment (IQA) metrics has become more pressing with the advent of new applications and services [1]. Objective image quality assessment can be divided into three categories [2]: Full-Reference (FR), Reduce-Reference (RR) and No-Reference (NR). Both FR and RR are essentially image similarity or fidelity measurement since they need the original undistorted image as a reference. However, in many practical applications, an image quality assessment system does not have access to the reference images. Therefore, the Blind Image Assessment (BIQA) which can evaluate image quality without using any reference image information, has become the focus of the research.
