Automatic registration method for laser point cloud and sequence panoramic image

Abstract

Aiming at the problems that the quality, the precision and the spatial data reference of unmanned aerial vehicle / vehicle-mounted mobile measurement laser point cloud and the sequence panoramic image data are not uniform, and registration errors exist among multi-source imaging data, the invention provides an automatic registration method for realizing the automatic registration of the laser point cloud and the sequence panoramic image based on the SFGM (Space Fingerprint Graph Matching). On the basis of registration primitive extraction, homonymy matching is achieved through a space fingerprint map method, then robust registration is achieved from coarse to fine, and the technical bottleneck of point cloud and image automatic data registration is broken through. The result of the invention lays a solid data foundation for the joint analysis and processing of multi-source spatial data, and serves the fields of basic surveying and mapping, smart cities, navigation, location services and the like.

Description

technical field [0001] The invention belongs to the application of automatic data registration in the fusion of unmanned aerial vehicle / vehicle mobile measurement data, and proposes a brand-new automatic registration method of laser point cloud and sequence panoramic image - SFGM (Space Fingerprint GraphMatching, spatial fingerprint map matching) matching standard method. [0002] technical background [0003] The UAV / Mobile Mapping System (Mobile Mapping System, mobile measurement system) integrating LiDAR (Light Detection and Ranging, laser ranging), panoramic camera and other multi-sensors provides a new technical means for remote sensing earth observation. The traditional photogrammetry and remote sensing data are expanded in terms of direction and data resolution, making it possible to obtain multi-angle, high-resolution, and multi-source data from low-altitude top surfaces and panoramas, street view sides and panoramas, and provide urban scene modeling and target extrac...

AI Technical Summary

Problems solved by technology

The invention proposes a SFGM (Space Fingerprint Graph Matching) registration method: study the robust extraction method of laser point cloud and sequence panoramic image registration primitives, generate heterogeneous geometric registration primitive sets; study graph-based laser The point cloud and sequence panoramic image conjugate registration primitive pair fuzzy matching method solves the difficulties of inaccurate geometric positioning and the search of primitives with the same name in the collection of non-injective heterogeneous registration primitives; and then studies the two-step high-precision registration The robust estimation method of model parameters, the self-conjugate primitive pair solves the initial value of the registration model, weakens the dependence of the model on high-precision POS data, and realizes the coarse registration parameters through the conversion of different-dimensional registration to the same-dimensional registration problem Refinement to achieve high-precision and robust registration of UAV / vehicle MMS laser point cloud and sequence panoramic images

Images