Multi-fused physical signature persistence clustering segmentation method

Abstract

The invention belongs to the field of computer graphics and topology, particularly provides a multi-fused physical signature persistence clustering segmentation method, and mainly overcomes the shortcomings of long running time, unstable segmentation result and relatively poor robustness of an existing segmentation method. The method comprises the steps of firstly, selecting required physical signatures; computing fused three physical signature functions; generating a persistence graph by utilizing persistence clustering; and finally, selecting a threshold value for combination, and generatinga segmentation result. The segmentation method is suitable for any three-dimensional mesh model; the segmentation result is good, and the segmentation speed is greatly increased; and the method has aremarkable effect on the fields of reverse engineering, medical imaging, model deformation, local matching and the like of three-dimensional mesh models.

Description

technical field [0001] The invention belongs to the field of computer graphics and topology, and specifically proposes a multi-fusion physical signature persistent clustering and segmentation method, which can be used for the feature description of grid models and the segmentation of various three-dimensional grid models. Background technique [0002] Three-dimensional grid segmentation is an important field in computer graphics, and it is a basic operation in geometric processing. In the description of grid feature points, concepts such as various curvatures or normal vector angles are mostly used as watershed functions, and curvature The features displayed by the angle with the normal vector are usually not robust enough. On the contrary, with the gradual deepening of research on thermonuclear signatures, wave nuclear signatures, local binary descriptors based on thermal diffusion, and scale-invariant thermonuclear signatures, their applications in 3D The field of model se...

AI Technical Summary

Problems solved by technology

[0003] There are many 3D mesh segmentation methods. For example, the method based on weak convex segmentation is not universal, and the segmentation results are different each time; the segmentation method based on Mean-Shift has large limitations in segmentation results and is only applicable to branch shape features. Objects that are not compact; the method based on consistency segmentation performs well for the segmentation of large models, but it is not suitable for smaller models; the segmentation method based on machine learning, although the segmentation effect is good, it takes too much time offline, which leads to the overall system Slightly bloated

Images