Research On Reconstruction Method Of High Fidelity Surface Triangulation Model

In recent years,with the advent of 3D laser scanning technology and oblique photogrammetric technique,meanwhile with the rapid development of consumer-grade drone,the ways of getting geographic data are deeply changed and become much more efficient.These specialities like speediness,high efficiency,low cost for getting spatial data basically provide much safer and more reliable data for the digital city construction,also reduce the cost of digital city modeling,speedup the period of modeling,and give a boost to the applying of point cloud data in the digital city construction.Based on the role of oblique photogrammetric technique playing in the digital city building,this paper will have a research on three-dimension surface reconstruction of building's point cloud and especially focus on the problems existed in current point cloud reconstructed methods,like losing lots of features of buildings'structure,and do not represent the original appearance of geographic object as comprehensive and life-like as possible.By use of images got through the measure process of oblique photogrammetric technique to assist the point cloud segmentation with its rich color,texture and other information.I just come up with a way to make blocking point cloud be constructed and jointed based on the surface growing method for the point cloud after segmentation,so that the completed reconstruction of surface modeling can be achieved.The following are the main research contents and results:(1)Based on the theory of pinhole imaging principle to research the process of transform between point cloud data and image data,to design methods of projecting and transform from three-dimension point cloud to two-dimension images,finally to make these 2 kinds of data integration and correlation be possible.Furthermore,provide a simple and efficient method via deeply analyzing point cloud data to solve the potential problems like repeated projection and penetration projection in the transform process.And finally do some experiments and analysis to check and test the efficiency of above mentioned transform method.(2)The point cloud segmentation method of integrated point cloud data and image information is studied,mature watershed image division algorithm technique is be used to segment the images correlated with point cloud data,then segment the point cloud according to the images segmentation results and relations between point cloud data and image pixel.Next reasonably solve the over-segmentation problems by taking advantage of some characteristics of point cloud data itself has.And the result of segmentation for blocking point cloud shows this method is much better than the method of simple segmentation for point cloud data.It can not only achieve the better segmentation for point cloud especially applying well to the construction mutation on the surface of buildings,but also maintain the construction features of geographic entities.(3)Do a research on the surface reconstruction and joint strategies for blocking point cloud,by use of prior surface growing rules to reconstruct and solve the mutation problems on the surface of building.Based on the position and topological relations between adjacent surfaces after reconstruction,jointing separately and make the edge mutation of object's surface reconstructed correctly.And the result of reconstruction indicates that the segmentation did on the position of feature mutation can effectively keep the mutation characteristics on the surface of object.Besides,it can reduce the problems like growing in a wrong direction in the position of object's feature mutation when the surface is growing and expanding.So we can draw a conclusion that this method can correct the topological relation between discrete points and reconstruct the surface well.