Research And Implementation On Fusion Of Airborne Laser Point Cloud And CCD Images

Airborne LIDAR technology is a very advanced technology, it has become animportant means to obtain the geospatial information, and widely used in urbanthree-dimensional modeling, coastal protection. Compared with photogrammetry, itsadvantage lies in short cycle time, low cost and high accuracy data, but the density pointcloud data acquired by LIDAR cannot show the real color of objects, which has difficulty inobjects classification and extraction. Whereas high-resolution CCD image data can providethe color of object, which is complementary to the point cloud. The fusion of these two datahas great significance in visualization and data processing.This paper first introduces the principle of point cloud and CCD images, and detailsthe CCD camera model. Then we designed two different fusion algorithms based on theanalysis and summary of fusion algorithm of the point cloud and the CCD image data.1）With the camera calibration, we can calculate the Euler angles sequence from the rotationmatrix which is getting form the motion analysis of the images. Then the Euler anglessequence will match with the attitude angle sequence obtained from the IMU. Next we willtake the output of IMU and GPS as the external camera parameters and use collinearequation to complete the fusion of point cloud and image.2）Without the camera calibration,the first step is image stitching. With the direct linear transformation algorithm (DLT) inphotogrammetry, using at least six corresponding points in point cloud and stitching image,the parameter of DLT equation can be solved. Finally we find out the RGB from the imageand put it to the corresponding point cloud to achieve the fusion.This paper mainly discusses the fusion of point cloud and image data. And theexperiment was carried on to verify the two algorithms. We can clearly identify the objectfrom the point cloud after the fusion with the image. The research above laid the foundationfor the data processing.