Naturalization of Screen Content Images for Enhanced Quality Evaluation

Abstract

SUMMARYThe quality assessment of screen content images (SCIs) has been attractive recently. Different from natural images, SCI is usually a mixture of picture and text. Traditional quality metrics are mainly designed for natural images, which do not fit well into the SCIs. Motivated by this, this letter presents a simple and effective method to naturalize SCIs, so that the traditional quality models can be applied for SCI quality prediction. Specifically, bicubic interpolation-based up-sampling is proposed to achieve this goal. Extensive experiments and comparisons demonstrate the effectiveness of the proposed method.

DIBR-Synthesized Image Quality Assessment via Statistics of Edge Intensity and Orientation

“…Higher values of PLCC, SRCC, KRCC and a lower value of RMSE indicate a better performance. They are calculated after the five-parameter logistic mapping [12]:…”