Method for extracting three-dimensional coordinate values from image in robot scene

Abstract

The invention relates to the technical field of coordinate value extraction, and discloses a method for extracting three-dimensional coordinate values from an image in a robot scene. The method comprises the following steps: performing layered modeling on XY plane images of a camera at different heights (Z) from a target surface, fixedly mounting the camera and a point laser on proper positions at the tail end of a robot, moving the robot and an additional shaft in a photographing posture to make a real-time image center cross line fall on a target point, and obtaining the a three-dimensional coordinate values of a certain point on any plane under the condition that the sight line of a camera is basically perpendicular to a target surface. The method is used for rapidly and accurately obtaining the three-dimensional coordinate value of any position point on the target image, and is beneficial for the robot to carry out operation path planning. The point taking precision is high, the welding precision requirement of the robot can be met, intelligent target feature path extraction can be conducted on a static image through the image processing technology, and the connected device can avoid sliding displacement in the coordinate positioning process due to excessive smoothness.

Description

technical field [0001] The invention relates to the technical field of extracting coordinate values, in particular to a method for extracting three-dimensional coordinate values ​​from images in a robot scene. Background technique [0002] Using a robot for welding work, quickly obtaining the actual path coordinates of the target weld is the key task. Using the teach pendant for interactive operation is far from satisfying the automatic continuous operation of the robot. It is easy to generate pixel coordinates in the image coordinate system from the world coordinate system, but Extracting pixel coordinates from the image coordinate system, and then performing a series of coordinate transformations to generate three-dimensional coordinate values ​​in the world coordinate system is not easy to achieve. This requirement is very urgent in robot application scenarios. If the target position can be quickly extracted The path coordinates make it possible to apply robot operations ...

AI Technical Summary

Problems solved by technology

[0003] Usually, a coordinate point in a three-dimensional world can be transformed by the robot transformation matrix to find a corresponding pixel point in the image, but it is very difficult to find its corresponding point in three dimensions through a point in the image. Existing The method used in the technology uses traditional images to obtain coordinate point values ​​with insufficient accuracy and low efficiency. Usually, it is difficult to obtain accurate Z-direction coordinates when obtaining image points. It is not easy to quickly convert image 2D coordinates into world 3D coordinates. Due to the size of the camera Limitation, the effective working range is fixed, it cannot adapt to large-area and large-size components, and it cannot meet the needs of some scenes that require high point-taking accuracy, such as welding

[0004] Usually, in the process of switching according to different points of the positioning camera, the outer shell and the inner shell slide against each other, and after multiple operations, the inner wall will cause excessive friction. Therefore, when positioning the coordinates, the side wall of the camera is prone to loosening and slipping. Due to the situation, there will be errors in the positioning coordinates. When the light switches, the sliding between the outer shell and the inner shell will easily cause the base to shake due to excessive force

Images