Description: The problem of creating an integrated full-reference visual quality metric based on neural networks that takes advantages of the best quality metrics is considered. The analysis of the effectiveness of existing metrics on complex types of distortions database test images TID2013. The criteria of choice of metrics incorporated in the integrated metric based on neural network, and using cross-correlation function held their selection. Developed integrated metrics based on various schemes of artificial neural networks. Using a database of test images TID2013 conducted verification of synthesized metrics visual image quality and selection of the optimal scheme. Validated best match human perception metrics integrated visual image quality in the presence of a reference.
Keywords: measures of visual quality in the presence of a reference, the system of visual perception of the person, image analysis, neural networks