Point cloud data-based primitive detection method and apparatus, and computer device

Abstract

The invention relates to a primitive detection method and device based on point cloud data and computer equipment. The method comprises the steps of inputting acquired sample point cloud data into a to-be-trained primitive detection model, determining a plurality of sample primitives corresponding to the sample point cloud data, and fitting geometric parameters corresponding to the sample primitives; through the to-be-trained primitive detection model, extracting a global structure relationship among a plurality of sample primitives according to the geometric parameters corresponding to the sample primitives; according to an error structure relationship in the global structure relationship, constraining the to-be-trained primitive detection model to obtain a pre-trained primitive detection model; and performing primitive detection on the obtained cloud data of the point to be detected through the pre-trained primitive detection model. By adopting the method, the primitive detection accuracy can be improved.

Description

technical field [0001] The present application relates to the technical field of artificial intelligence, and in particular, to a primitive detection method, device, computer equipment, storage medium and computer program product based on point cloud data. Background technique [0002] The point cloud data scanned by various 3D sensors (such as Kinect, LiDAR, etc.) can save the space, color, intensity and other information of the object and the surrounding environment, but it is not convenient for people to directly understand and analyze. Moreover, since the point cloud data itself is sparse, noisy and disordered, it is impossible to obtain an accurate high-level abstract semantics directly from the point cloud data. That is, there is an understanding gap between the low-level visual features expressed by point cloud data and the high-level abstract semantics that people can handle. By detecting ubiquitous geometric primitives in point cloud data, one can build bridges acr...

AI Technical Summary

Problems solved by technology

[0003] However, the traditional method uses an isolated primitive detection method. When the structure of the man-made object is complex, detection errors are prone to occur, resulting in low primitive detection accuracy.

Images