A research on method for classification of Vehicle-borne laser colored point cloud data

As a new means of data acquisition, vehicle-borne laser scanning technology can quickly obtain accurate the three-dimensional spatial information of objects, it is very suitable for urban objects in the rapid access of three-dimensional spatial information. The colored point cloud data by fusing laser data and CCD image, not only having spatial information, but also the color attribute information. In this paper, colored point cloud data is the object of the study, the author proposed a strategy of stepwise classification: projection grid classification combine with supervised classification. Firstly, all the data points are projected on the horizontal grid, according to the properties of grid cells information, they are divided into three categories: ground class, building class, and other object class. Secondly, for each category, using the method of supervised classification, according to spectral information (RGB), dividing the points into different classes in order to achieve the classification and extraction of surface features. Experimental results show that different kinds of objects can be extracted and classified effectively using the method.