Depth Image Segmentation And Compression

The three-dimensional reconstruction model is widely used in the fields such as pattern identification,cultural relic reconstruction, digital amusement, city planning,CAD and reverse engineering. Analysis of range data is the basis of the following three-dimensional reconstruction.On the basis of studying and analyzing obtaining principle and method of range data, with the significant scientific research of Beijing: service system of three-dimensional spatial information of Beijing and the plan of academic innovative team of Beijing :obtaining and processing technology of three-dimensional spatial information,the dissertation focuses on noise removal, segmentation and compression of range data . The major contribution of this dissertation can be summarized as follows:1. making comparative analysis of the noise removal methods of range data, providing a good data preparation for follow-up treatment.2. putting forward new segmentation method of range image based on combination of RHT and surface curvature.3. bringning forword compression method of range image based on stratified sampling of surface curvature. Not only retaining local details of the object, but also reducing volume of points data. Compressing rate can be controlled according to actual needs.In conclusion, by plentiful experiment and programme simulating, the research succeeded in segmenting and compressing range data, providing to a certian extent technology accumulation and preparative for homegrown development of laser three-dimensional data process.