Stacked scattered target point cloud segmentation method based on convex region growth

Abstract

The invention provides a convex region growth-based stacked scattered target point cloud segmentation method, and the method mainly comprises the following steps of carrying out smoothing processing on stacked scattered target point cloud acquired by a depth camera by adopting a moving least square algorithm; solving normal vector and curvature information of all points in the smoothed point cloud based on a principal component analysis method; and finally, clustering according to the normal vector, curvature and concavity and convexity characteristics of the point cloud to realize point cloud segmentation. According to the invention, the point cloud segmentation algorithm only depends on the features of each point in the point cloud, can achieve a good segmentation effect of the stacked scattered target point cloud without pre-training the target model, has high operation efficiency, and has good adaptability and robustness for point cloud segmentation in different scenes.

Description

technical field [0001] The invention belongs to the technical field of three-dimensional point cloud processing, and in particular relates to a point cloud segmentation method for stacked scattered objects based on convex region growth. Background technique [0002] With the continuous popularization of depth cameras and the continuous development of 3D vision technology, 3D point cloud processing technology is playing an increasingly important role in the fields of robot intelligent sorting and 3D vision inspection. For the intelligent sorting tasks of robots that randomly stack workpieces in industrial sites or randomly place objects in life scenes, the 3D point cloud data of the target acquired by the depth camera may have uneven density, unstructured data, etc., and the target point cloud The mutual adhesion between objects is not conducive to subsequent processing, and the scattered stacking of objects makes the task of point cloud segmentation more difficult. [0003]...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the defects and problems existing in the existing methods, the present invention proposes a point cloud segmentation method of stacked scattered objects based on convex region growth, which solves the problems of low operating efficiency and poor segmentation effect existing in the existing point cloud segmentation methods of stacked scattered objects. Ideal problem, in the present invention, the normal vector and curvature estimation of the smoothed point cloud are obtained, more accurate normal vector and curvature information are obtained, and then clustering is performed according to the normal vector, curvature and concave-convex characteristics between neighboring points , compared with the point cloud segmentation method based on region growing, the concave-convexity judgment condition is added, which avoids the situation of over-segmentation and under-segmentation of the point cloud. The element clustering improves the point cloud segmentation efficiency, and the method of the present invention can achieve a relatively ideal segmentation effect under the premise of ensuring the segmentation efficiency

Images