Research Of 3D Visualization Based On The Fusion Of Laser Point Clouds And Image Data

3D laser scanner can easily obtain the point clouds data of the target surface.But the raw point clouds data contains only spatial geometry information,Meanwhile,High resolution CCD camera can get 2D color image of the target scene.The image data can provide real color texture information of the target scene.The point clouds data and image data can be combined to build the most realistic virtual 3D model,which has a wide prospect in many 3d visualization fields such as digital city,virtual reality.The research of 3d visualization base on the fusion of point clouds and image data mainly involves three key problems: point clouds and image data registration,3D visualization of point clouds,the fusion of point clouds and multi-view images.First,Inspired by the calibration method of 2D laser scanner and camera.A method of registration based on Point-to-Plane ICP is proposed,which uses the RANSAC algorithm to extract the plane calibration plate from point clouds.The optimal transformation between the two sets of data is obtained by optimizing the distance of the corresponding plane in laser scanner coordinate system and the camera coordinate system,and then the transformation between the point clouds and the image data is calculated.Secondly,For the 3D visualization of the point clouds,An up-sampling method based on the result of the MLS surface projection is adopted.The resolution of point clouds and utilization of pixel were increased in the premise of preserving the original characteristics of the point clouds,which enhanced the visual effect of colorized point clouds.Afterwards,A multi-resolution octree data structure is created to deal with large data sets.Powered by WebGL,the colorized point clouds was rendered over the web browser,which expand the accessibility and usability of point cloud data.Finally,the multi-view images is obtained by rotating the camera,and the estimation accuracy of the camera rotation parameter is improved by image feature matching,which solves the registration of point cloud data and multi-view image data effectively.Based on the research above,a 3Dvisualization hardware and software system based on the fusion of laser point clouds and image data is implemented,which realizes the functions of raw point clouds and image data acquisition,data preprocessing and fusion,and 3D visualization based on WebGL.