Research On Fram Rate Up Conversion For Stereoscopic Video Based On Motion And Depth Information

In recent years, stereoscopic film and stereoscopic television begin to come into people’s life. Stereoscopic videos offer strong and intense visual impact. It can give the audience rich visual experience. Because the volume of stereoscopic video is very large, we often reduce the frames to reduce the time resolution in the process of transmission and storage. Based on the correlation information between the original video sequence frames, it can improve the time resolution and enhance the visual quality of stereoscopic video.This thesis researchs the motion estimation and motion vector optimization of frame rate up conversion for stereoscopic video, and combine the saliency algorithm with the video segmentation algorithm. The thesis combines the depth information of stereoscopic video with motion information effectively, and realizes the frame rate up conversion of stereoscopic video. The main work of this thesis is as follows:(1) Research the basic principle and method of frame rate up conversion. In this thesis, a variety of motion estimation methods, motion vector processing and motion compensation methods are researched.(2) Combine the method of frame rate up conversion with saliency algorithm. Firstly, the dual motion estimation algorithm is studied, and the depth information constraint is introduced to improve the accuracy of motion estimation. Then I research the saliency detection algorithm based on spatial and temporal, and propose a saliency detection of frame rate up conversion for stereoscopic video. In view of the salient region and non salient region, different methods are used to optimize the motion vectors.(3) Combine the method of frame rate up conversion with video segmentation algorithm, and propose a method of motion estimation based on video segmentation for stereoscopic video. Based on the stereoscopic video segmentation method of meanshift, we apply the segmentation graph information and the consistency energy function and the gradient information to the block matching process of motion estimation, so that it can improve the accuracy of motion vector and the visual quality of the interpolated frame.This thesis is based on the platform of MATLAB. The method of frame rate up conversion with saliency algorithm for stereoscopic video has been proposed. The method of frame rate up conversion with video segmentation algorithm for stereoscopic video has been proposed. Multiple stereoscopic video sequences have been tested. The parameters about PSNR and SSIM have been improved. The test results show that the two methods are effective and the quality of interpolated frames has been improved.