Surface difference visual detection method

Abstract

The invention discloses a surface difference visual detection method which comprises the following steps: collecting a surface image of a detected object, carrying out light bar center extraction andimage processing on the image to obtain a light bar image, and carrying out contour segmentation on the light bar image to obtain a left light bar contour and a right light bar contour; obtaining coordinates of end points at the two ends of the single-side light bar contour, connecting the two end points, traversing all points on the single light bar contour, searching for a point farthest from the connecting line, marking the point as a demarcation point, dividing the light bar contour into two parts by the demarcation point, and marking the part far away from the other light bar as an outercontour; obtaining the outer contour of the light bar on the other side by adopting the same method; selecting a reference contour, measuring the contour, and performing point cloud fitting on the reference contour to obtain a reference line. According to the method, the surface difference is calculated by analyzing the point cloud of the outline of the measured object, so that the measurement efficiency is effectively improved.

Description

technical field [0001] The invention relates to the field of visual detection, in particular to a visual detection method for surface difference. Background technique [0002] Surface difference is a common form in industrial processing. Whether the size of surface difference is accurate or not is related to the aesthetics and sealing of the entire workpiece. It is a very important parameter in quality monitoring. For example, in the process of automobile production, the size error of the cover Larger surface differences of the body will lead to larger surface differences, which will affect the appearance, dryness, and sealing of the vehicle. Therefore, in the manufacturing process, it is necessary to measure the surface difference of the vehicle body in real time; This method has low efficiency, low precision, and is easily affected by the subjective factors of the measurement personnel, so it cannot meet the high-efficiency, high-paced and high-precision measurement requir...

AI Technical Summary

Problems solved by technology

[0002] Surface difference is a common form in industrial processing. Whether the size of surface difference is accurate or not is related to the aesthetics and sealing of the entire workpiece. It is a very important parameter in quality monitoring. For example, in the process of automobile production, the size error of the cover Larger surface differences of the body will lead to larger surface differences, which will affect the appearance, dryness, and sealing of the vehicle. Therefore, in the manufacturing process, it is necessary to measure the surface difference of the vehicle body in real time; This method has low efficiency, low precision, and is easily affected by the subjective factors of the measurement personnel, so it cannot meet the high-efficiency, high-paced and high-precision measurement requirements; with the development of computer vision detection, visual recovery based on line structured light The technology of three-dimensional profile measurement has attracted more and more attention. The line structured light sensor is composed of a line structured light projector and a camera. It actively projects the line structured light onto the object to be measured. Fluctuations, etc.) to determine the size parameters of the measured object, such as in the surface difference image, the light bar usually appears convex and disconnected (the surface difference position of the measured object has a gap) or obviously convex (the position only has surface difference ) situation; the collected structured light feature points are restored to the three-dimensional space coordinates of the camera coordinate system. In this way, the surface difference characteristics of the measured object can be obtained more efficiently and accurately

Images