A hybrid no-reference video quality assessment based on region of interest

In this paper, a hybrid no-reference video quality assessment based on region of interest (ROI) is proposed. Firstly, two types of ROIs, motion area and skin region are detected and integrated as the final ROI in each frame. Secondly, a blockiness measurement based on ROI in pixel domain is introduced. Then, the blurriness is measured in frequency domain, by using two-dimensional discrete cosine transform (2-D DCT). At last, the assessment result is combined with the blockiness and blurriness value. By calculating the correlation between the assessment result and the subjective score (mean opinion score, MOS) in different definition videos, the experiments show the advantages of the proposed method over some other methods.