Extraction Algorithms on Weld's Defect Based on Digital Subtraction Angiography Algorithms

摘要

In order to improve the detection efficiency of nondestructive testing of weld defect ray and reduce labor intensity of operation, an intelligent detection and extraction method for weld defects based on the digital subtraction angiography (DSA) algorithms is introduced, making full use of the information of weld images, according to the common difficult problems of accurately extracting the ray automatic detection. Based on the method of automatically choosing threshold, fuzzy set is used to transform gray degree in order to enhance image. Pixel labeling method is utilized to extract weld and keep its original grey value. Digital filter method is used to remove defects of weld on the original image, which is set as background image. And then, weld defect can be got according to DSA method, after which characteristic parameters of defects can be calculated finally. The results show that this method can accurately test the boundaries and defects of all kinds of weld images, which has a good adaptability and practicability.