Research Classification Based On Typical Vehicle For Rapid Prototyping Laser Point Cloud City

With the continuous development of the information society and the "Digital Earth" and "digital city" proposed urban surface extraction and application of information is becoming increasingly important. Vehicle laser measurement system to quickly and accurately obtain information obtained three-dimensional space features makes it a quick solution to urban spatial information collection and data updates a good choice. But for quick access to the city’s three-dimensional model and three-dimensional information, the classification and rapid prototyping automotive laser point cloud is its essential step.For the laser point cloud processing ultimate goal is modeled, and in order to meet the efficiency, the best is fast to automatic configuration of network modeling, rapid prototyping and the foundation for the classification of the point cloud.There are no more systematic and sophisticated classification methods specifically for laser scanning systems and vehicle laser point cloud for rapid prototyping of classification. For this problem, this paper draws on some of the more sophisticated airborne laser data processing experience, and laser scanning data based on vehicle characteristics, presents a vehicle for rapid prototyping laser point cloud classification methods, in order to get the city feature the critical information and services in rapid prototyping automotive laser point cloud. Which the contents of this paper are:1) Acquisition and pre-processing of the raw data, analyzing the characteristics of the laser data and vehicle data and airborne data because of differences in the performance characteristics of the data which led to the difference between the two categories, according to the text and to determine the classification method.2) Classification of vehicle laser data. Features of Car laser scanning data, and the city’s main features. To the performance characteristics of surface features, the use of this article from the definition of the parameters and shape characteristics were different for different feature extraction methods:the ground, the use of seed points in the original data surface fitting method, the ground is extracted, for lights, trees and buildings, the main part of such an object on a vertical surface, in accordance with the first height level, shape and characteristics of each target using the extracted first and differentiate, and then the key information of each target (lights:the height and position, trees:canopy, tree diameter, tree height, location, buildings:height, edge line, the ground:the grid surface after thinning) with different file storage.3) Read the previous step classification results, which were stored in a different object to read the key information stored file, choose a good modeling tools and modeling methods, and then import the three-dimensional scene, to generate three-dimensional modeling of urban scenes.As can be seen from the experimental results of this paper, the proposed algorithm for vehicle laser point cloud classification can be in good effect and high accuracy. Classification result can be directly used for rapid modeling of individual targets, has a very important practical value.