Feather point matching method based on colored false random coding projection

Abstract

The invention belongs to the filed of image matching of coding structured light active visual sense in machine visual sense, in particular to a feather point matching method based on the colored false random coding projection. The invention is based on a false random coding principle, and designs a novel colored coding projection template by using a point and line combined method. The method comprises the following concrete steps: 1. generating a false random sequence from a0, a1 to an by a feedback network n-displacement bit register specified by a primitive polynomial h(x)=xm+hm-1xm-1+...+h1x+h0; 2. filling the false random sequence from a0, a1 to an into a matrix with the size of n=n1*n2 for generating a false random array b; and 3. using the generated array values as the discrete feather points, connecting the feather points by a feather line, and establishing a coding template. The invention introduces a morphological algorithm, and provides the extraction algorithm of the relevant feather points. In the decoding process, the invention provides the thought of mutual verification of adjacent windows, can realize the automatic extraction and matching of the feature points, and can perfectly solve the problem of shadow generation during the projection of the structural light on the surface of the complicated three-dimensional scenes and the problem of mistake feather point matching caused by that parts of coding patterns are covered because the shooting angles of a video camera are different.

Description

technical field [0001] The invention belongs to the field of image matching of coded structured light active vision in machine vision, and relates to an active vision technology based on the principle of color pseudo-random coding combined with coded structured light illumination, and specifically relates to a feature point matching method based on color pseudo-random coded projection . Background technique [0002] A very important goal of machine vision is to recover the 3D coordinate data of the scene from the 2D image and reconstruct the 3D model of the 3D scene or object, even if the computer has the ability to recognize the 3D environment information through the 2D image information. It has been widely used in robot positioning and navigation, reverse engineering, object recognition, virtual reality, product monitoring and other fields and has achieved good social and economic benefits. [0003] In the process of 3D visual scene reconstruction, a well-known problem is...

AI Technical Summary

Problems solved by technology

[0009] The present invention firstly designs a coding projection template based on the principle of pseudo-random coding. The template adopts the combination of feature points and feature lines, and each feature point is relatively discrete, and is connected with each other through feature lines, which can overcome the existing templates because of the problem of coding. The defect that the isolation or connection of color units will hinder decoding; in the decoding process, a method of mutual verification of adjacent windows is proposed, combined with algorithms such as opening and closing, erosion and expansion in mathematical morphology, and additional algorithms are given to achieve Automatic extraction and matching of all feature points eliminates the problem of matching errors in complex scenes

Images