No-Reference Quality Assessment for Screen Content Images Based on Hybrid Region Features Fusion

Linru Zheng,Liquan Shen,Jianan Chen,P. An,Jun Luo

DOI: https://doi.org/10.1109/TMM.2019.2894939

IF: 7.3

2019-01-23

IEEE Transactions on Multimedia

Abstract:Research on screen content images (SCIs) attracts more attention as they are highly applied to image- and video-centric applications on mobile and other devices. It is important to develop an efficient image-quality assessment (IQA) method for SCIs because IQA can guide and optimize various image-processing methods for SCIs and improve user experience. In this paper, we propose a no-reference objective assessment model for SCIs including SCIs segmentation and the analysis of local and global perceptual feature representations. Since the human visual system is highly sensitive to sharp edges that are commonly encountered in SCIs, we utilize the variance of local standard deviation, which is a noise robust index to distinguish the sharp edge patches (SEPes) and non-SEPes of SCIs. For SEPes, we perform two kinds of feature extractions. First, the entropy and contrast features are extracted with a gray-level co-occurrence matrix, which are highly perceptive of microstructural change. Second, the local phase coherence is utilized to capture the loss in sharpness. Then, average pooling is adopted to fuse features obtained from all of the SEPes to represent the local features. We further combine local features with global features that are derived using the BRISQUE method as the hybrid region (HR)-based features. Finally, a regression module is learned using support vector regression to train the mapping function that maps HR-based features to subjective quality scores. Experimental results on the screen image-quality assessment database show that the proposed method can achieve better performance in visual-quality prediction for SCIs than the performance achieved by state-of-the-art methods.

Computer Science