S-CIELAB model and iCAM model are used for estimating the image quality by calculating the color difference between the original image and the distorted image. When the models is used to evaluate the image difference, the S-CIELAB model compensates for the effect of spatial scale of the human visual system. iCAM model, in addition to considering the spatial scale dependency, takes into account the effect of chromatic adaptation on colour appearance. The effectiveness and accuracy of these two kinds of models are investigated based on visual perception experiments using three images with sharp colour contrasts. The study shows that the predictions of iCAM model are more consistent with human visual performance than those of S-CIELAB model.