Stereo Video Quality Assessment Based On Motion And Disparity Information

Because of the stereoscopic vision can give people a realistic immersed sense, such as the virtual represent of tourism landscape and the 3D display of Cultural relics, etc. The Visual shock Stereo video brought about, makes people have not merely satisfied to the HD video and prompts the flat TV to stereo TV reform. In order to seize the preemptive future of digital television commanding point, the key technologies of Stereo video, become the research hot spot both at home and abroad. How to obtain high quality stereo video after processing is the key factor to three-dimensional imaging technology move toward the market key factors. So how to determine the three-dimensional video has better quality, need the work of stereo video quality evaluation, which is a challenge at present in the 3D field.This paper firstly proposes a video quality evaluation method to the single viewpoint which based on gradient and motion estimation (GMES) according to the existing evaluations and human visual system. The experiments results show that it is coherent to the human eye’s visual effects on the distortion coded videos.3D video have an increase of parallax information compared to the single viewpoint video, so this paper start the evaluation research of stereoscopic video mainly from two sides: quality and stereoscopic perception. As to quality evaluation of the binocular stereo video, the paper put the proposed method into the quality evaluation of each channel, and educed the results. Then according to the human brain properties for stereoscopic perception, the final video sequence quality index is obtained by using subsection weighting to the objective quality evaluation results of left and right channels. In the three-dimensional sense side, a disparity estimation algorithm for stereo video embodiment of three-dimensional sense extraction was proposed. Firstly, obtain disparity map of each frame both reference video sequence and distorted video sequence; Secondly, use the SSIM method on each disparity map to calculate the similarity comparison both reference and the distorted video sequence; Finally, the test video sequence stereoscopic perception index of is obtained by weighted average of the disparity map similarity comparison results.In order to verify the effectiveness of the proposed method, the paper uses Pearson correlation coefficient method to measure the correlation between the two indicators and the subjective evaluation results respectively. The experiments results show that it is coherent to the human eye’s visual effects and the correlation coefficient achieved reached above 80%.The method provides a technical scheme in the stereoscopic information quality.