Abstract: We present a new full-reference image quality metric using Gaussian-Hermite moments and human visual system (HVS). Orthogonal moments are powerful tools in pattern recognition and image processing applications. Low-order moments well characterize features of an image and can be used to assess the image quality accurately. Firstly the proposed method obtains the feature of low-order moments through computing the continuous orthogonal moment energy differences between two images. Then, by combining the masking effect of human visual system (HVS), the different weighting factors are assigned to different areas in an image. Finally, image quality scores are derived by the weighted average of the continuous orthogonal moment energy differences in all the areas of the image. The performance of the proposed method is evaluated on several public databases. Experimental results and comparisons demonstrate the efficiency of the method.

Key words: Gaussian-Hermite moments, image quality assessment, the orthogonal moment energy differences, human visual system, masking effect