The Building Extraction From Airborne LIDAR Point Cloud Data Based On Strip Division

As the expand usage of the spatial data and the increment of people's requirements, fast and accurate information extraction for urban buildings become more and more urgent. The traditional mapping technologies, whether in speed or accuracy can not meet people's needs. Laser scanning (LIDAR) technology to solve this problem appears to offer hope for people.Laser radar technology is a new mapping technology, which integrates a laser rangefinder (Laser), high-precision inertial measurement unit (INS) and global positioning system (GPS).It is a high technology, not only can get high-resolution rate digital surface model quickly, making people free from the heavy data collection, but the system itself and the features of point cloud make the data-processing towards the automation direction further. In recent years, laser radar technology has developed rapidly:from 1995 to today, its market share in the mapping has incremented from 5% to 12%,7.1% growth every year.LIDAR technology is already mature in the hardware, there have been many commercial products today, but the development of data processing software technology is a little slower, seriously affecting the application of laser radar technology.The main objective of this paper is the information extraction of urban buildings, primarily to find a measure to extract the information of the buildings'outline from the LIDAR point cloud data collected by the LIDAR system, focused on the thought of strip division in the processing of LIDAR data. The main contents of this paper are listed as below:(1)Research the method of LIDAR data processing using strip division strategy. Strip division strategy mixes the advantages of the ideological of divide and conquer and the thought of dimension reduction, not only makes the large LIDAR point cloud processing becomes possible, but also can reduce the complexity of the algorithm largely to shorten the running time of the algorithm.(2)Research the application of divide-and-conquer triangulation in the LIDAR data processing. First triangulates LIDAR point cloud data using divide-and-conquer method, and then separate the building segment, extract the edge point set of each segment.(3)Research the applications of random sample consensus algorithm (RANSAC) in the LIDAR data processing. RANSAC algorithm contains a large number of knowledge of probability and statistics, it can extract the mathematical model from the given data set. In this paper, the traditional RANSAC is improved so that it can adapt to a given data set that contains several mathematical models.(4)Research the method of line adjustment according to the characteristics of buildings. The lines which extracted using RANSAC algorithm are not fully consistent with the characteristics of the building. In this paper, these lines are divided into four categories, and then each category is adjusted separately.The results of experiment show that the strip division strategy is completely feasible for the LIDAR point cloud data processing. It can get a good compromise effect between the time cost and the effect of the processing result, so the strip division strategy is very suitable for the building information extraction from the large-scale LIDAR point cloud.