Vehicle door appearance detection method based on machine vision

Abstract

The invention provides a vehicle door appearance detection method based on machine vision, and belongs to the field of appearance detection methods, and the method comprises the steps: copying a filtered vehicle door welding appearance image, and setting the images before and after copying as an image 1 and an image 2; zooming the image 1, inputting the zoomed image 1 into a neural network A, outputting an image defect confidence coefficient, comparing the image defect confidence coefficient with a set threshold value A, counting and marking welding defects, and grading the welding defects to obtain a welding defect grading result; and performing color channel conversion on the image 2, performing segmentation to obtain image blocks, extracting welding area features, normalizing the welding area features into color features, inputting the color features into a neural network B, outputting an image non-welding probability, comparing the image non-welding probability with a set threshold B, performing statistics and marking on non-welding areas, and performing grading B on the non-welding areas to obtain a welding integrity grading result. Basic processing and target detection are carried out on the vehicle door appearance image to obtain a welding defect and welding integrity analysis result, so that the detection becomes faster and more efficient.

Description

technical field [0001] The invention relates to the field of appearance detection methods, in particular to a machine vision-based car door appearance detection method. Background technique [0002] In the welding process of automobile doors, various uncertain factors may cause missing welding and uneven welding of the doors. In the post-welding inspection of car doors, manual visual inspection is slow in inspection speed, prone to false inspections and missed inspections, low in efficiency, and unable to meet production requirements. In the batch appearance inspection of car door welding, it is necessary to quickly and efficiently detect and judge welding defects and welding integrity in order to shorten the automobile production cycle. In order to solve the disadvantages brought by the manual visual inspection method, a new type of detection method is needed, combined with modern machine vision technology, to detect and analyze the appearance of the detected car door and ...

AI Technical Summary

Problems solved by technology

[0002] In the welding process of automobile doors, various uncertain factors may cause missing welding and uneven welding of the door.

In the inspection of the door after welding, the detection speed is slow by manual visual inspection, and it is prone to false detection and missed detection, which is inefficient and cannot meet the production requirements.

Images