Distortion types identification based on singular value decomposition and BP neural network

In this paper, we proposed a neural network approach to identify distortion type of images in image quality assessment, in order to remove subjective factors to make the quality evaluation much more fair. Based on image singular value decomposition, the algorithm mixed with neural networks for training images, reflecting image quality qualitatively. Repeated experiment results show that the method has advantages of simple, fast and high efficient. For the loss compression (jpeg, jpeg 2000), white noise and Gaussian blur, the method also has good results in image distortion type identification.