Vision real-time measuring method based on color code block

Abstract

The invention discloses a vision real-time measuring method based on a color code block, comprising the following steps of: S1, arranging a color code block with a given size above the back of a robot; S2, collecting a color code image of the color code block by using a single camera; S3, tracking the central point, the middle point of a long side and the middle point of a short side of the colorcode block to determine the region of a color code block image; S4, scanning and processing a color code image according to the set interval and direction to obtain an edge point with sub-pixel precision; S5, carrying out least square fitting on the edge point with sub-pixel precision of the color code block to obtain four edge straight lines; S6, solving points of intersections of the four edge straight lines to obtain image coordinates of four vertexes; S7, calculating a new central point, the middle point of a new long side and the middle point of a new short side of the color code block; S8, solving the pose of the robot by utilizing an n-point perspective algorithm; and S9, optimizing the pose of the robot by utilizing an orthogonal iteration algorithm to obtain the pose with orthogonal pose matrixes of the robot.

Description

technical field [0001] The invention belongs to the visual measurement in the field of robots, in particular to a method for robot pose measurement. Background technique [0002] Currently, commonly used visual measurement methods mainly include binocular stereo vision and monocular vision based on an n-point perspective (Perspective-n-Point, PnP) algorithm. Because binocular stereo vision needs to process two images and feature point matching for each measurement, its real-time performance is poor. Monocular vision only processes one image per measurement, and its real-time performance is higher than that of binocular stereo vision (Xu De, Tan Min, Li Yuan, Robot Vision Measurement and Control, National Defense Industry Press, p. 130, p. 140 page, 2008). [0003] When the rectangular color-coded block with known size is used as the mark of the robot, the extraction of feature points is the extraction of the vertices of the rectangular color-coded block. At present, the c...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the technical problem that the accuracy of the feature points obtained by using intersection points is not high in the prior art, which affects the accuracy of visual measurement based on the PnP method, and the attitude matrix in the measurement results cannot be guaranteed to be orthogonal, the purpose of the present invention is to provide A real-time visual measurement method based on color-coded blocks for high-precision real-time visual measurement of robot pose and guaranteeing the orthogonality of the attitude matrix in the measurement results

Images