Full-Reference Contrast-Distortion Image Quality Assessment Algorithm


  • Hasan Thabit Rashid Kurmasha


Histogram Equalization HE, Distortions, Contrast Enhancement, Image Quality Measures, Human Visual Perception


Image enhancement is a popular technique used for process transferring distortions; contrast-distorted images CDI are treated using different images quality assessment IQA algorithms as full-reference FR-IQA. In this paper, a new FR-IQA is proposed to determine the noise artifacts caused by compression, blurring and other types of distortion as well as the brightness saturation (i.e. no details in the region) which happen for the same reasons above. The proposed FR-IQA is designated based on the variance moment which has the feature of fast computation, very sensitive to represent the dispersion of image pixel intensity (change in details), and very sensitive to the overall brightness of images in correlation to the human visual sensitive HVS. The statistical evaluation outcomes expression that most of the FR-IQAs have humble correlation along with human mean opinion scores MOS although they are using the original image in comparing with the distortion one. The proposed algorithm takes into account the significant properties of human visual perception HVP. The algorithm objective and subjective results have very well scores. It has significantly well correlation for person correlation coefficient PCC > 0.87, spearman rank order SROCC > 0.91 and outlier ratio (OR) equal to 0%.