Research On The Simplification Of Center Of Gravity In Point Cloud Region Based On Octree

The 3D laser scanning system is one of the main methods of spatial data acquisition.It can continuously acquire a large amount of 3D point cloud data on the surface of a scanned object.However,with the rapid development of high-precision and high-density point clouds,point cloud redundancy has also been brought about.In order to meet the actual needs of point cloud modeling with different precisions and to improve the speed of subsequent processing of point clouds,as few as possible point clouds are used to express more detailed object characteristics,the point cloud is reduced and compressed to become a large-scale point cloud data processing.It is a very necessary operation.On the basis of the research on the current point cloud compression method,the characteristics of point cloud data obtained by various systems are analyzed,aiming at the limitations of the classical Barycenter of area data compression method,combined with the characteristics of point cloud data and the three-dimensional space segmentation theory,it proposes a data-driven simplified algorithm for the point cloud Barycenter of area data within a leafnode based on adaptive linear octree partitioning.First,under the condition of data drive and compression ratio control,an adaptive octree spatial structure is established step by step.In the first step,according to the initial resolution and depth of the octree determined by the improved compression ratio correlation formula,the outermost bounding box space determined by the point cloud data is preliminarily divided,and then the point cloud is inserted into the initial octree.In the second step,according to the number of point clouds in the minimum bounding box after insertion,the threshold value range is determined under the control of the compression ratio,and the point cloud is adaptively partitioned.Then,based on the adaptive octree,traversal nodes establish a linear octree storage structure.The linear octree only stores the characteristics of the real-leaf nodes,which can reduce the memory space occupied when the depth of the tree is deep,increase the space utilization,and also improve the efficiency of the point cloud simplification of the saved real-leaf nodes.Finally,according to the same theory of the point cloud data attribute content and the same kind of object intensity information,the concept of “center of gravity” is expanded,and intensity information is added to the streamlined indicator.Using normalized thought would do to deal with dimensionless distance difference and intensity difference,and the data is comprehensively processed to determine the most appropriate voxel center of gravity.Using multi-source data to compare the effects of different compression methods under each level of compression ratio control,we find that step-by-step adaptive linear octree structure compression is more time-consuming than evenly divided three-dimensional raster voxel array compression significantly reduced,greatly improving the compression efficiency.At the same time,point cloud data collected in various forms still has excellent reduction effects at lower compression ratios.It also shows thatmulti-detailed level bounding boxes formed by adaptive octree-modified space division enlarge the versatility of the method,making it scattered point clouds can be applied to complex scenes.Based on the data index analysis of the standard deviation,the octree-based point cloud Barycenter of area improvement method proposed in this paper has a better compression effect on the point cloud than the traditional Barycenter of area data compression method under the same compression rate.