Research And Design On 3-D Reconstruction Environment System Based On Laser Radar Data

3-D environment reconstruction based on the data of laser radar, which is an important subject in the research and applications of optical 3-D sensing, is very important for promotion of optical 3-D sensing. This method can overcome the shortcoming of machine vision algorithm which the scene reconstructed is not unique. It also can provide accurate information for robots navigation and improve the ability of human-computer interaction as well. It is propitious of mobile robot navigation in unknown environment.Through analyzing of data properties of the laser radar, this paper investigates thoroughly the theories and techniques of 3-D environment reconstruction used laser radar. The main content is divided into four parts: data collection and handling, range data registration, reconstruction of triangle grid surface and grid fusion. The research content and fruit are provided as followings:A)A 2-D matrix can be obtained through laser radar scanning the environment around . Some measure errors will occur inevitably in the process. Aim at the noise problem, this paper use the idea of 2-D image filter algorithm,White noises and Gaussian noises can be reduced by using the image median filter algorithm. As for the traditional algorithm, repetitious comparison problem cannot be ignored. This paper presents a fast median filter after investigating the weighted median filter and choose plot filter. The speed and the effect of filter have been improved to some certain extent.B) The traditional ICP algorithm is adopted in this paper to register all range data collected from different viewpoints. Every range image should be aligned in a common frame of reference. In order to improve the speed of registration, the collection method of registration point pair has been developed. After the false pair of points being canceled, the number of point pair decreases. The run time is shorten as well. Adopting quaternion expression, this paper has computed the rotation parameter R and the translation parameter t to realized registration.C) Considering the inherent structural information exists in laserradar data, the shortest diagonal rule is chosen to divide the triangle gird. After those steps being finished, the triangle grid curve is produced successfully.In the process of image fusion, the redundant data in multi-angle image regulated should be eliminated. Then the images will be patched. It includes four steps as follows: determining the overlap region, deleting every viewpoint overlap region, determining patching edge-line, and fusing the grid through the patch algorithm.D) With the realization of the above methods, a 3-D environment reconstruction system based on laser radar can been exploited. A 3-D scene reconstruction can be accomplished when the camera can't identify clearly. So the purpose of 3 dimension reconstruction in the moving process of mobile robot has been achieved.