Active infrared CCD weld seam detecting method

Abstract

The invention is a kind of initiative infrared CCD welded seam detecting method. The infrared illuminant diode is used as the initiative infrared imaging light source, according to the micro organization structure and chemical ingredients change, acquires the welded seam initiative infrared image, acquires the video input of CCD with computer, then carries on a series of process to the welded seam infrared image, extracts character parameter to estimate the welding quality after image filter, image separation and image matching, and memorizes the result into data bank, then carries on statistic process to the welded seam quality, forms a report table to the fault seam.

Description

Technical field: [0001] The invention relates to an active infrared CCD welding seam detection method, which is used for welding seam detection in the industrial field, and is particularly suitable for welding seam detection of automobile parts, and belongs to the technical field of welding seam detection. Background technique: [0002] Due to the importance of weld inspection in the industry, people have conducted a lot of research work on weld inspection methods for many years, and now weld inspection has become a fast-developing technology. At present, there are mainly the following four effective methods: 1. Hard-ray (χ, γ) detection; 2. Ultrasonic detection; 3. Magnetic flux method detection; 4. CCD (solid-state camera) detection. The χ and γ rays in the hard-ray detection are harmful to the human body and are not safe. And it has to build lead houses to carry out the work. The cost of this kind of lead house is very high, and at the same time its other supporting equipment ...

AI Technical Summary

Problems solved by technology

Under visible light, the illumination model of the image is unstable and dynamic, so the image processing of the weld is relatively complicated, and the segmentation effect of the weld image is very poor, which affects the extraction of weld parameters

If intelligent learning methods such as neural network support vector learning methods are used, a certain segmentation effect may be achieved, but the speed is slow and not suitable for real-time detection.

Images