Quality difference ranking model for smartphone camera photo quality assessment

Image quality assessment (IQA) for smartphone cameras is important in mobile phone performance evaluation, but existing image quality assessment methods are not suitable for smartphone camera photos. In this work, we propose a quality difference ranking model for smartphone camera photo quality assessment, which compares and ranks image quality from four distortion types. First, we select some image patches in a pair of distorted images according to the image content and distortion types. Then we input these image patches into the quality difference ranking model to predict the quality order of the two images. Finally, we use the bubble sort to rank the quality of all images. Experimental results show that our model outperforms previous IQA methods in smartphone cameras photo quality database.