Registration method of laser 3D point cloud and CT 3D point cloud

Abstract

The invention discloses a registration method of laser three-dimensional point cloud and CT three-dimensional point cloud, comprising: performing statistical filtering, point cloud region growth processing and voxel filtering on the patient's body surface laser three-dimensional point cloud data, and combining the patient's 4D-CT The data is reconstructed into CT 3D point cloud data; the point set S is selected from the laser 3D point cloud data 1 , select point set S from CT 3D point cloud data 2 ; Using the PointRegNet network, from S 1 Extract feature point set Q, from S 2 Extract the feature point set V; set all the feature points in Q to be connected to each other to form a set E of feature edges S , set all the feature points in V to connect with the feature points of K neighbors to form a set E of feature edges, and connect the graph (V, E) and the graph (Q, E S ) for registration; use the motion amplitude information to S 1 Assign weights to the points in S 2 Based on the establishment of the GMM model, the GMM model combines the weighted S 1 The points in construct the probabilistic likelihood function. The technical scheme of the invention alleviates the problem that the prior art cannot be directly applied to body surface point cloud registration.

Description

technical field [0001] The invention relates to the field of medical point cloud registration, in particular to a registration method for laser three-dimensional point cloud and CT three-dimensional point cloud. Background technique [0002] The rapid registration of laser 3D point cloud and CT 3D point cloud is a hot topic in current research, and it has important clinical application value in the positioning of patients in radiotherapy and the registration of robotic surgical navigation systems. Based on the CT image data made before or during the operation, the location of the lesion in the patient's CT coordinate system is extracted, and the patient's CT three-dimensional point cloud data is extracted. Through the laser 3D imaging system mounted on the robotic arm, the patient's laser 3D point cloud data in the coordinate system of the robotic arm can be scanned. By registering the 3D point cloud scanned by the robot arm with the CT 3D point cloud data, the spatial posi...

AI Technical Summary

Problems solved by technology

However, due to the relatively flat surface of the human body and the influence of respiratory motion, the problem of accurate registration of 3D point clouds on the body surface is not a strict rigid registration problem.

Therefore, traditional point cloud registration methods cannot be directly applied to the registration of body surface point clouds.

[0005] In view of this, the object of the present invention is to provide a registration method for laser three-dimensional point cloud and CT three-dimensional point cloud, to alleviate the problem that the prior art cannot be directly applied to body surface point cloud registration

Images