| Stereo image and stereo video contain the depth information of the scene, which can provide viewers with the benefits of added realism, can be widely used in the fields of virtual reality, digital entertainment, remote control etc. Compared with single-view video, stereo video has much greater data amount, it is difficult to storage and transmission, so efficient compression is very important. This dissertation considers the problem in stereo video system include display technology, disparity estimation, mode decision, color correction.Firstly, fully discusses the basic theory of three-dimensional(3D) vision and the access of stereo video, focuses on some representative 3D display method. And analyses the common stereo image format and its mapping relationship with the between sub-screen. On this base stereo display software was developed, which could supply stereo image and stereo video sequence.Secondly, on the base of the stereo video coding framework proposed by 3DAV, motion compensate prediction(MCP) combination with disparity compensate prediction(DCP), implements a block-based stereo video coding in this thesis. The experimental results show that the scheme can remove the information redundancy between the views of the stereo video effectively. Meantime, takes advantage of the supplemental enhancement information(SEI) makes the stereo video bitstream compatible with the standard H.264 bitstream, implements the scalable views.Finally, according to the characteristic of stereo video, global disparity vector method is employed, and it was used in the mode decision, the compression efficiency improved significantly. Because of the different directions of stereo video camera, the received light intensity differences, at the same time the camera gain and electricity level not exactly the same, lead the difference color between the inter-view images, which will affect the performance of stereo video compression seriously. In order to improve the efficiency of stereo video compression further, based on the analysis of the color difference between inter-view images, this thesis implements a color correction algorithm based on histogram matching. |
PDF Full Text Request