Edge detection method and device based on grid data, medium and equipment

Abstract

The invention discloses a 3D disordered point cloud edge detection method and device based on a 5D texture grid data structure, equipment and a medium. The method comprises the following steps: (1) converting 3D disordered point cloud data into the 5D texture grid data structure; (2) carrying out neighborhood access: for any vertex Smn in the 5D texture grid data structure, acquiring all neighborhood points NS taking Smn as the center and having a radius of r; (3) carrying out Gaussian blur, assigning the RGB value of each vertex Smn as the RGB value weighted average of all neighborhood pointsNS corresponding to the vertex Smn; (4) obtaining a gray scale gradient gi and a gradient direction di of the 5D texture grid data structure after Gaussian blur; (5) performing non-maximum suppression on the gradient image, and outputting a numerical value Ti corresponding to Smn; and (6) searching the edge contour of the quadrangular grid M and outputting the edge contour, and converting the disordered point cloud into a 5D texture grid data structure, so that the disordered point cloud also has a rapid neighborhood access capability, and the edge detection of 3D disordered point cloud datais achieved.

Description

technical field [0001] The invention belongs to the technical field of computer graphics and computer vision, and in particular relates to an edge detection method of disordered point cloud data. Background technique [0002] In the field of machine vision, visual features can be applied to scenes such as object recognition, pose estimation, SLAM, etc., where the edge profile of an object, such as the texture profile of a surface pattern, is a commonly used visual feature. For the edge detection of 2D images, there are mature algorithms, such as Canny operator, but 2D images cannot fully represent 3D objects. The 3D point cloud contains more three-dimensional pose information of the object, and the edge detection for the 3D point cloud has a wider application value. [0003] The edge detection method of 3D point cloud generally includes the following parts: (1) blurring the original point cloud of the object acquired by the 3D camera; (2) using the RGB value of the point cl...

AI Technical Summary

Problems solved by technology

[0004] The above 3D point cloud edge detection method is a generalization of the 2D image edge detection method, but it is only applicable to ordered point clouds

3D point cloud is divided into two types: ordered point cloud and disordered point cloud. Ordered point cloud is a 3D point cloud collected by a 3D camera in the order of rows and columns. It is usually filtered to remove invalid points. Point cloud The order is difficult to guarantee; in addition, many commonly used 3D vision algorithms, such as the following sampling, are easy to destroy the order of ordered point clouds, and then convert them into disordered point clouds

Unordered point cloud is a more common form of 3D point cloud, and mature 2D edge detection methods are difficult to extend to unordered point cloud

Images