Segmentation Methods Of Points Cloud Data Based On Images

Since the concept of ’Digital Earth’ has been proposed, the research on the acquisition of the spatial information has become a hotspot. 3D laser scanning technology is a kind of spatial information acquisition technology that is highly developed, and it’s called ’Real Copy Technology’,because the advantages of this technology, such as high measuring precision, time saving, strong sense of reality. Recently, 3D laser scanning technology is widely used cause its technology getting much more mature and the reducing of the scanner cost, so its application become more. Points cloud data acquainted from 3D laser scanning technology is analyzed by scholars all around the world, and this made points cloud data become a important data source among ’Digital Earth’, 3D model building and virtual reality.Points cloud segmentation is based on the extraction of spatial geometric information, at the same time, it’s the basement of point cloud data’s feature extraction and analysis. Also, the segmentation is not that easy cause points cloud data is normally discrete and non-structured. The results of further data processing is determined by the efficiency and accuracy of the segmentation. So the research on point cloud segmentation is very meaningful to us.This article firstly introduced the 3D laser scanning technology, and then emphatically discussed about the fusion of points cloud data and image data. Then established the relationship between digital image and points cloud data of the same station with the direct linear transformation method from photogrammetry, and realized the fusion of point cloud data and image data. After that we segmented the image with interactive image segmentation technology and mapped the results of image segmentation to the point cloud data based on the relationship between image and point cloud data. With all these we can realize the region of interest points cloud segmentation.With the points cloud data from VZ-1000 of Reigl scanner and the corresponding region’s image data from camera, we verified the correctness and accuracy of algorithm by realized the fusion of points cloud data and image data though programming. Then we segmented the image data and points cloud data and experiment shows that we can realize the complex texture region’s points cloud segmentation by using interactive segmentation based on the fusion of image data and points cloud data.