Abstract: In order to evaluate image quality efficiently and objectively，this paper proposes a Curvelet transform and grey relation analysis based method which utilizes the global comparison mechanism of grey relational analysis theory and the analysis ability in multi-scales and multi-direction of Curvelet transform.Firstly，the grey relational grades between evaluating images and the reference image at different scales and directions are gained.Secondly，the mean of the relational grades at all angles in the same scale is computed，and then，a global relational grade between the mean and the reference sequence is produced which provides us with an image quality in two levels.Experimental results show that the new method not only may estimate image quality at different scales and angles，but also produces more reasonable conclusions than PSNR based methods.