No-Reference Stereoscopic Image Quality Algorithm Combined With Binocular Vision Characteristics

Stereo images are affected by distortion in the process of image acquisition,coding and transmission,and the degree of distortion affects the quality of received images,which can also reflect the performance of image processing system.Therefore,how to design the stereoscopic image quality assessment(SIQA)algorithm that conforms to human binocular vision perception is of great significance.In this paper,several kinds of SIQA algorithms are briefly introduced.What's more,on this basis,the SIQA method combined with binocular vision characteristics is deeply studied.We proposed the corresponding stereopaires assessment models,the specific content as follows:1.A SIQA algorithm based on DCT domain features is proposed.Synthesize the left and right images to two “cyclopean” maps respectively according to the binocular fusion and the binocular competitive characteristics.Extract the statistical features in DCT domain by using normalized DCT coefficient of the distorted image pairs and the obtained two “cyclopean” maps..Four image features are fused as a feature vector of stereopairs.Support vector regression(SVR)is used to to predict the objective scores of stereoscopic images by establishing a map between the stereoscopic image feature vector and the difference mean opinion score(DMOS).The experimental results show the proposed method is consistent with subjective perception,and hence outperform the existing methods.2.A SIQA algorithm based on DCT domain and spatial domain features is proposed.In order to extract image perception features more comprehensively,the proposed combines DCT domain and spatial domain features of stereopairs.The experimental results show that the performance of the algorithm has been improved by combining spatial features,which verifies the effectiveness of combining different domain features.