| Stereo systems are designed to emulate human stereo perception, which have many applications. One major obstacle for the application of stereo videos is the extremely large amount of data associated with a stereo sequence. There are mainly two methods to compress stereo video, block-based coding and object-based coding. Block-based coding has the merits of simplicity and robustness, but may cause blocking artifacts. Object-based coding is very desirable as it can avoid blocking artifacts, and has higher coding efficiency. Furthermore, it can describe a scene in a structural way. Therefore, it has been an active area of research as the developing trend of stereo video compression scheme.Video object segmentation is a first crucial step in object-based stereo video coding. The video object is the area in the image, which has certain physical, visual and semantic meaning. In the study of images, images are split into several regions. The certain regions which draw more attention of researchers are called objects or foreground, and the rest regions are called background. We can use different compression methods and compression ratio to deal with different objects, so that we could get good quality. As a whole, there are many video object segmentation methods, which have own strongpoint and shortcoming. There isn't a universal segmentation algorithm for all kinds of images. Therefore, object segmentation in stereo is a problem full of challenge.This paper mainly works on the object segmentation problem in stereo videos. For stereo images, an object segmentation algorithm based on disparity map is proposed. A stereo video segmentation algorithm combining disparity map segmentation and multi frame difference segmentation is also proposed in this paper. The algorithm combines the strong points of disparity map with twice frame difference. First, segment disparity maps to get the object segmentation template in different disparity lays. Then make twice frame difference extraction in the template regions and use edge extraction to modify the object edge. After that the precise motion object has been gotten. Over segmentation can be avoided in the case that there is motion interaction between moving objects or motion information are involved in the background.After object segmentation, carry out comparison on three stereoscopic video coding schemes by the experiments. That the left channel use ordinary objects of MPEG-4 encoding, and the right channel object use in two ways: one is based on the right channel MCP previous frame image, the other is based on the left channel image corresponding frame DCP way, and then choose from a smaller forecast errors. MCP can not predict movement from the "exposed" areas (the former one is exposed can not be seen in part of the campaign, in the next exposed in the frame), the DCP can not predict "shielding" regional. MCP and DCP will be combined; they will make up for shortfalls, thereby reducing forecast error. The main task of this paper is as follows:1.First, analysis of various stereo matching methods and existing images, video segmentation methods have been made. Then, a new stereo video segmentation method based on disparity map and frame difference is proposed. At the beginning, stereo map is under-Sampling to reduce the computation of stereo matching. Second, disparity map is got by matching a one-dimensional window in the horizontal scan line of stereo video under-Sampled. Then object of different layers of the disparity map is first segmented, and over-sampled to get a fist segmentation template. Third, a further matching computation is made in the region of the first segmentation to get a more precise disparity map. Finally, combing with results of edge detection and multiple frame, the final results of the segmentation come out.2. Stereo video coder based on object is implemented which mainly consists of shape coding, texture coding and motion estimation and compensation. The left access is the main sequence and the right access is the supplemented sequence. Simple MPEG-4 encoding is using in the left access, motion and disparity compensation is made in the right access. The best match block is searched by entire search in the criteria of absolute error. And then obtain the corresponding residual, motion vector and disparity vector. Find the smallest residual by comparing respective residuals, storage corresponding motion or disparity vector. Stereo image is realized by integer transform, quantization and entropy coding based on the context.3. After adaptation of the official MPEG-4 standard model, a new coding algorithm of stereo video based on MPEG-4 standard is realized. Comparison of several video coding by experiment, object-based video coding has better compression results; and good reconstructed images quality are obtained by experiments on sequences of different image complexity. So the algorithm is effectiveness.Through detecting the stereo video, it can be seen that the stereo vision object segmentation algorithm proposed in this paper can obtain accurate motion object. In addition, the algorithm is of low computational complexity and easy to realize. The algorithm is effective on segmentation motion object of stereo video.The segmentation of stereo images is efficient for stereo images with small occlusions. When there are large occlusions, complicated occlusions extraction and disparity compensation must be taken. Our work is based on gray information of images and videos. In future work we anticipate that by involving color information, it should be possible to get more accurate segmentation results. At the same time, the object-based coding of stereoscopic is only on the model of MPEG-4.So, there are some points which need to improve further. |
