Blind Image Quality Assessment by Fast Quality Assessment Network

Abstract

In this paper, a new neural network structure, which is called fast quality assessment network (FQA-Net), is proposed for fast blind image quality assessment (BIQA). FQA-Net is a very simple neural network, which mainly includes convolution layer, standard deviation measurement layer and regression layer. In order to improve the efficiency of the network, a group of visual filters (VFs) are obtained by simulating the neurons in the cerebral cortex. The VFs are used as the convolution kernels in the convolution layer, then the outputs of the convolutional layer are a set of feature maps. After that the standard deviation of each feature map is calculated directly. Finally, the regression function is used for the mapping between the standard deviation values and the quality scores. FQA-Net not only reduces the number of parameters and the output dimensions in the training process, but also prevents network overfitting effectively. The experiment results show that FQA-Net has relatively low computational complexity and high competitiveness compared with the leading BIQA methods.