Target recognition method based on feature geometry returns

Abstract

The invention in particular relates to a target recognition method based on feature geometry returns. The problems that results such as key point detection, feature matching, mismatching and the likein a conventional three-dimensional point cloud target recognition process are non-ideal, target recognition effects in complex scenes with noise, different resolution ratios and existence of barrieroverlapping and the like are poor and the like are solved. The target recognition method comprises the following steps: increasing a step of removing edge key points at a key point detection stage; eliminating a feature matching pair with ambiguity at a feature matching stage by utilizing a nearest neighbor ratio; providing a feature geometry return method at a hypothesis-generation stage, gathering accurate matches, and generating hypothesis transformation; finally, accurately verifying hypothesis transformation at a hypothesis verification stage, and accurately estimating the posture of thetarget. The method is applicable to recognition of messy three-dimensional point cloud scene target models and applied to target recognition of machine vision. The method can be widely applied to thefield of unmanned vehicle driving, robots, automated assembly, intelligent monitoring and the like.

Description

technical field [0001] The invention belongs to the technical field of machine vision, and in particular relates to a target recognition method based on feature geometric returns. This method is suitable for the recognition of target models in messy 3D point cloud scenes, and is used for target recognition in machine vision. Background technique [0002] Vision is an important means for human beings to perceive and understand the world. Computer vision technology allows computers to acquire, process, analyze and identify images by simulating human vision to realize the understanding of the real world. Object recognition has always been one of the research hotspots in the field of computer vision, and can be widely used in areas such as unmanned vehicle driving, robotics, automated assembly, and intelligent monitoring. The purpose of target recognition is to identify the target of interest from the scene and obtain its attitude information such as position and orientation. ...

AI Technical Summary

Problems solved by technology

[0004] In view of the unsatisfactory results of key point detection, feature matching and mismatch elimination in the existing 3D point cloud target recognition process, and the poor effect of target recognition in complex scenes containing noise, different resolutions, and overlapping occlusions, etc., The present invention proposes a method for target recognition based on feature geometric returns

Images