Research Of Three-dimensional Laser Scan Data Compression

With the computer and three-dimensional imaging's rapid development technology, three-dimensional digital technology gets people's attention and application more and more. Generally three-dimensional digital technology can be divided into data acquisition, Surface reconstruction and surface information acquisition process. Surface grid reconstruction in three-dimensional objects from the mass point cloud data generates the correct topological relationship of objects' grid, which is the bridge of point cloud data and the surface model. Three-dimensional reconstruction of the surface grid is directly related to the quality of the objects cognitive, feature extraction and three-dimensional measurement applications, such as follow-up. On the other hand, the general three-dimensional surface reconstruction process is in the context that point cloud data are known but the relationship of objects' topology is unknown, which involving in computational geometry, topology, as well as various fields such as computer graphics knowledge. Thus three-dimensional surface reconstruction algorithm is of great theoretical and practical significance.Through the laser scanner to obtain three-dimensional point cloud data, because of various errors and the existence of noise, the original data will result in irregular, difficult to surface reconstruction. Therefore, it is necessary to carry out three-dimensional pre-processing the original data. After the objects are scanned the large number's original point cloud data of a great density will generate, and in which noise are randomly distributing. When using these data to complete the grid, surface reconstruction, three-dimensional reconstruction and display, it is important to carry out effective data pre-processing in order to ensure the accuracy of the data collection. In scanning process, stray light, background light, or system operating and a number of other accident factors had resulted in some data which are not belong to its own , which has also been recorded in the data file. Therefore, before laser scanning's the data points to obtain renewal the data must be carried out to remove noise.(1) Data pre-processing. At the base on the PCA of wavelet de-noising, threshold de-noising method, I have deal with the point cloud and analysis the results. (2) Three-dimensional surface reconstruction. I have optimized the three-dimensional grid by local optimization algorithm to the basis lines of mesh optimization. It can reduce the number of grids while maintaining quality of grid showing, and improve the display speed.(3) I have written the PCA wavelet de-noising and threshold de-noising program at the Matlab platform. And I reconstructed different three-dimensional point cloud data, obtain good results.