A Method for Recognition of Defects in Welding Lines

Defects detection and recognition is very important to guarantee the welding quality. Many researches have done on this topic. However, most of them are too complex or limited in efficiency. Different from these researches, a method that is used to auto-recognize defects from the images of welding lines is put forward. Begin with the characteristics of the welding defects, the paper discusses an effective method to extract the features of the defects much simple algorithm which is based on perceptron model to recognize and classify the defects. Experimental results show that the pretreatment of the images of welding lines is very important to the feature extraction and defects recognition and the and method of recognition and classification of defects put forward is effective. It can much reduce the working effort of human being and increase the defect recognition efficiency.