Abstract: This paper proposed a image duality assessment algorithm based on wavelet transform. Here, both reference and test images were decomposed into four bands: LL, HL, LH and HH. The edge structure information similarity was obtained on other three bands, then we got a universal image quality assessment result(CSSIM) with weights based on the human property. The proposed metric performs better than MSSIM and MUSSIM by providing larger correlation coefficients and smaller errors after nonlinear regression fitting.

Key words: Structural similarity, Wavelet transform, HVS