Feature fusion based no reference quality assessment for screen content image

Considering the human visual system is more sensitive to the edge and local texture information, this paper presents a feature fusion based no reference quality assessment model for screen content image (SCI). In the proposed method, the Histogram of Oriented Gradient and Local Binary Pattern are exploited to describe the edge and local texture information of the SCI. A feature fusion process is subsequently conducted to obtain a feature to better reflect the distortion of SCI. Finally, the support vector regression is applied to obtain the quality assessment mapping model from the above fused feature to subjective rating. Experimental results show that the proposed method is able to better reflect the human perception on SCI, compared with multiple state-of-the-art image quality assessment methods.