3D feature extraction method and system based on TOLDI descriptor and medium

Abstract

The invention discloses a 3D feature extraction method and system based on TOLDI descriptors and a medium. The method comprises the steps that three new planes are drawn at the positions parallel to three coordinate planes of a pre-established local coordinate system and with the distance being the supporting radius r, and all neighbor nodes around a feature point p in a three-dimensional object are projected to the three new planes respectively; dividing a limited projection area into a plurality of square grids on a projection plane; counting the projection distance from each neighbor node in the grids to the plane on the plurality of square grids, and selecting the point with the smallest projection distance as the representative point of the plurality of square grids; and extracting features of the feature points according to the projection distance from the representative point to the plane and the included angle of the z axis z (p) in the local coordinate system. According to themethod, the repeatability of the improved local coordinate system in a common scene is higher, a better effect can be achieved under some shielding conditions, and the wrong matching condition in thematching process is prevented.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a TOLDI descriptor-based 3D feature extraction method, system and medium. Background technique [0002] The features of a 3D object refer to information that can represent a 3D object, such as some geometric information and spatial information, and its features are generally divided into two categories: global features and local features. [0003] In a three-dimensional object, the extraction of local features is divided into two steps: first, the points with rich information are detected as feature points, and then the features of the feature points are extracted. The feature extraction is based on the geometric information and space around the feature points. Distribution information, because the feature is extracted at each point, so this method is very robust to occlusion, and the extracted features are also very descriptive. The local features used in this ...

AI Technical Summary

Problems solved by technology

[0009] Yang et al. proposed a Triple Orthogonal Local Depth Images (TOLDI) in 2017. When establishing the local coordinate system, in addition to using the covariance matrix, the descriptor also uses surrounding neighbor points The spatial distribution of the coordinate axis is used to establish the coordinate axis, so that the coordinate system has high stability in complex scenes, but the establishment of the local coordinate system of the TOLDI descriptor makes it very repeatable in complex scenes, because in When establishing the z-axis, only a small part of the neighbor points is taken to establish the covariance matrix

The purpose of this method is to reduce the influence of occlusion, but at the same time it will also make the information used when establishing the z-axis too scarce. The lack of information may make the gap between the established z-axis and the real one too large, especially in some areas without occlusion On the object, it is equivalent to wasting a lot of information in vain

Images