Multi-Indicator Image Quality Assessment of Smartphone Camera Based on Human Subjective Behavior And Perception

摘要: As smartphones are widely used in daily lives, manufacturers and customers increasingly concerned about the performance of smartphone cameras. However, due to unique distortion types, there relatively few image quality assessment (IQA) methods for images. In this paper, we propose a photo model, which scores from four aspects: color, texture, noise exposure. Based on human observation behaviors different indicators, two novel cropping viz. SalGAN-crop based saliency prediction SSIM-crop structural similarity proposed. Different features after-wards extracted by simulating subjective perception, predicted finally given AdaBoost regression analysis. Experimental results reveal that our model can provide more accurate than traditional methods.