New image quality metric using random projection

This study proposes a new image quality metric using random projection (RP). Generally, distortion is wide spread over the entire region in an image when the image is degraded by compression, blurring, additive noise and so on, which signifies that it is not necessary to compute the degree of distortion at every pixel. RP has been used for representing a signal to a small number of measurements. In this study, to evaluate image quality, distortions between the reference and distorted images are measured at a small number of pixels that are preselected or randomly chosen through the RP and to beat the simple random sampling, the saliency-based weighted sum is used. Experimental results with the laboratory for image and video engineering data set show the effectiveness of the proposed quality metric.