Three-dimensional scanning system circular index point detection method

Abstract

The invention relates to a round mark point detection method of a three-dimensional scanning system. The method firstly performs the pre-treatment to an image taken by a left and right video camera, the mark point is characterized in that a boundary in an outline map obtained after pre-treating is continuously closed, secondly, the boundary continuous closed area is taken as the characteristic of a mark mode, the corresponding characteristic detecting method is designed to detect the possible mark point in the image, the possible mark point is grouped into a suspected point set, and then the space corresponding point of each suspected point is utilized to set up the shape description a threshold in a self adapting way to the location in the point cloud and an angle between a micro-section plane and an imaging plane, and the threshold is utilized to perform the verification to the practical shape description result of the suspected point, to obtain the mark point in the image; at last, the matching is performed to the mark point in the image taken by the left and right video camera to obtain a mark point pair corresponded on the left and the right, and the space mark point corresponding to the matching mark point is obtained.

Description

Technical field: [0001] The invention relates to an object detection and recognition method, in particular to a circular marker point detection method in a three-dimensional scanning system, belonging to the technical fields of image processing and computer vision. Background technique: [0002] In the three-dimensional measurement system, the grating pattern is projected onto the surface of the measured object, and the left and right cameras installed in different directions are used to capture the grating fringe image of the object. For image processing, through the established camera mathematical model, the three-dimensional surface contour or three-dimensional space point coordinates of the object can be calculated according to the grating stripe image information captured by the left and right cameras. [0003] For large surfaces, due to the limitation of the measurement range of the measurement system, it cannot be measured completely at one time, and must be measured ...

AI Technical Summary

Problems solved by technology

In the actual three-dimensional measurement system landmark detection, because the circular landmarks have small area and random distribution, they have little influence on the gray distribution of the image, so it is difficult to select an appropriate segmentation threshold; The actual three-dimensional shape is deformed, and the circle is deformed into an ellipse after projection. The commonly used Hough transform method is not good for detecting non-standard circles with small areas, and the calculation amount is relatively large, so it is difficult to achieve real-time detection; The deformation in the image after point projection is also different. Selecting a single fixed threshold to verify possible marker points is prone to misdetection, and it is necessary to adaptively set the threshold for each possible marker point for verification.

Images