|
In the field of industrial inspection, because of a variety of problems that arise in the welding process, welding can lead to contain pore and crack defects and so on, affect the product’s quality and safety. So the welding image defect detection is very important. The influence of the traditional X-ray weld image of testing technology and experience, weld defect detection gradually transition to the computer intelligent identification from artificial detection. Using the X-ray real-time detection system of on-line detection and analysis, can effectively overcome the errors caused by artificial detection, thus make online testing work objecification, standardized.This topic research on portable digital weld film detector, the foundation of the digital image processing technology, the design has the characteristics of portable image acquisition and processing system. This paper discusses the portable image acquisition and processing system development, research, and technical realization. And the weld image recognition system is established based on the BP neural network by the application of Matlab, and faulting type of pattern recognition.Firstly it introduces digitized weld film development situation of detector at home and abroad, then introduces the overall design of image collection system. System is divided into four parts:most agencies, stepping control section, image acquisition and system, and the lighting design and functions in detail, including the design, preached agencies micro controller, system design and firmware program of stepping motor speed control with characteristics. Paper also introduces image preprocessing, edge detection and identification of weld defect plate algorithm and so on.According to the characteristics of the Euclidean distance metric judgment basis, determine the defect aspect ratio, circularity, Equivalent area, the gray scale difference between defects and the background, defects in their gray deviation,and the relative position of defects. use the method of BP neural network to weld defects in the film pattern recognition.
|