New Objective Stereo Image Quality Metric Using Human Visual Characteristics and Phase Congruency

Objective quality metric evaluates image quality automatically. In this paper, a new objective stereo image quality metric is proposed with left-right image quality (LRIQ) model and stereo perception quality (SPQ) model. In LRIQ model, wavelet transform is used to simulate multi-channel effect. Meanwhile, phase congruency (PC) map of sub-image is extracted as main feature to measure quality of each sub-image. Then all the sub-images qualities are weighted according to contrast sensitivity function curve. The SPQ model, quality score is obtained by comparing the PC maps of original and distorted absolute disparity images. Finally, these two models are combined to evaluate stereo image quality. Experimental results demonstrate that the correlation coefficients between proposed evaluation method and DMOS are above 0.93, and the root mean square errors are all less than 5.6, under JPEG, JPEG2000 compression, Gaussian blurring, Gaussian white noise and H.264 coding distortion. It indicates that the subjective results perform highly accordance with objective qualities.