Single-camera multi-view calibration method based on robot

Abstract

The invention provides a single-camera multi-view calibration method based on a robot. The single-camera multi-view calibration method comprises the following steps: S1, calibrating the robot and a camera; s2, obtaining corresponding points of the robot and the camera; s3, transforming the corresponding points to obtain a transformation matrix T; s4, the camera shoots the workpiece at different shooting points; s5, selecting a photographing point as an original point; s6, acquiring a translation vector Ti of the camera at other photographing points relative to the original point through the robot; s7, multiplying the translation vector by the transformation matrix to obtain a translation vector taking a pixel as a unit; and S8, unifying coordinate systems to realize single-camera multi-view calibration. According to the single-camera multi-view calibration method based on the robot, coordinate systems are unified, under the condition that the view fields of shot objects do not coincide, a single camera is used for conducting measurement, positioning, scene splicing and the like on a workpiece, and the problem that multiple cameras are needed during multi-view calibration is solved.

Description

technical field [0001] The present invention relates to the technical field of robot vision, and more specifically, to a robot-based single-camera multi-view calibration method. Background technique [0002] With the increasing application of machine vision, the demand for camera calibration in multiple fields of view is getting higher and higher. The use of a single camera can reduce costs, and the coordinate system of a single camera with different fields of view can be applied to large-format high-precision positioning and measurement And scene stitching and other applications. The difficulty of calibrating a single camera in multiple fields of view is that it is impossible to use traditional methods to stitch images in a unified coordinate system when each field of view does not overlap. [0003] In the prior art, multiple cameras are used to calibrate in different fields of view, and the cost is relatively high. For example, the high-precision large-field-of-view machi...

AI Technical Summary

Problems solved by technology

The difficulty of calibrating a single camera in multiple fields of view is that it is impossible to use traditional methods to stitch images in a unified coordinate system when each field of view does not overlap.

[0003] In the prior art, multiple cameras are used to calibrate the inventions and creations in different fields of view, and the cost is relatively high. For example, the high-precision large-field-of-view machine vision measurement and calibration device and method with the publication number CN109099883A are used for high-precision large-scale The measurement and calibration of the field of view is to shoot with multiple cameras, and then stitch the shooting fields of multiple cameras into a large field of view, which is expensive and not suitable for large-scale promotion

Images