Laser point cloud and dense matching point cloud fusion method

Abstract

The invention discloses a laser point cloud and dense matching point cloud fusion method, which aims at solving the problems of quality degradation and layering existing in fusion of ground laser scanning point cloud and aerial photography dense matching point cloud, carries out data redundancy removal and quality improvement on mixed point cloud on the basis of air-ground heterogenous point cloud registration, and discloses the laser point cloud and dense matching point cloud fusion method. The method comprises the following steps: firstly, constructing an energy function data item according to a distance between a dense point cloud and a laser point cloud and a normal vector included angle, constructing a smooth item by utilizing a geometric neighborhood relationship and color difference in the dense point cloud, and optimizing by adopting a graph cut algorithm to obtain a dichotomy label set of the dense point cloud; removing an overlapped redundant region between the two types of point clouds according to the labels; and finally, according to a neighborhood point selection strategy, carrying out surface curvature weighted guide point cloud filtering on dense matching points near the boundary, and combining different-source point clouds to obtain a fused point cloud.

Description

technical field [0001] The invention relates to a heterogeneous point cloud fusion algorithm in three-dimensional reconstruction of urban scenes, belonging to the field of remote sensing surveying and mapping. Background technique [0002] Point cloud is an important data source for 3D digital model reconstruction. In order to obtain the point cloud of the urban scene surface, there are currently two main types of measurement technologies, namely laser scanning method or structured light scanning method based on active vision, motion recovery structure based on passive matching algorithm, and multi-view stereo vision algorithm. LiDAR scanning technology is widely used in urban scene reconstruction, but limited by the field of view and occlusion of the sensor, a single scan or even multiple scans of the terrestrial laser scanning system cannot guarantee the integrity of the point cloud model, which is usually missing in the scanned point cloud. Building roof point cloud. Th...

AI Technical Summary

Problems solved by technology

On the basis of point cloud registration, some works have proposed some methods to improve the accuracy and integrity of the stitching point cloud, such as using the gradual migration method to achieve smooth connection of the border of the stitching point cloud, but the point cloud near the gap with a large distance cannot Move effectively; or use the multi-viewpoint projection method to detect holes in the laser point cloud, extract the corresponding data from the registered dense matching point cloud to fill the hole, and use the Laplacian fusion method based on the differential domain to enhance the accuracy of the merged point cloud Surface details, but the algorithm cannot achieve point cloud fusion outside the boundary of the laser point cloud

Due to the complexity and huge size of the directly mixed scene point cloud, the above method fails to smooth the mixed boundary of the heterogeneous point cloud under the premise of maintaining the uniqueness of the mixed point cloud data so as to achieve the result of point cloud precision fusion

Images