Ultrasonic Testing Of Image Restoration Techniques Based On Independent Component Analysis Theory

Recently, with the development of integrated circuit and computertechnology, image detection technique of digitization, automation and intelligentis becoming a hot issue. A lot of nondestructive test imaging technology havebeen used in industral producting. Independent component analysis (ICA) is anew blind separation of sources testing technique. It has been so important insome subjects. ICA technique has the theory and using prospects in wirelesscommunication, sonar, voice processing, image restoration and biomedicinefields.This paper studied the ultrasonic test C-scan image of brazing butt-weldedjoints which used in copper coil of lage motor or high speed electric locomotivetraction motor. This kind of joint must suit for the use of high power and smallsize, so the materal should has a character of high conductivity and high strength,therefore we choosed the copper alloy. As we know, weld and test processing isthe important technical for electric maching making. Ultrasonic immersiontesting was used as the nondestructive test method. Because of the non fuse is thefateful default for conductivity of coil, we designed and made a simulationsample with defect in it and after ultrasonic test the man made defct has beenfinded in the C-scan image. This paper summarizes several kinds of ICAcalculating theory and common methods. The main important point of this paperis that we studied the electric machine copper alloy brazing butt-welded jointrestoration image by ICA. We have a research in image noise reduction, edgedetection, image segmentation and image feature extraction. Finally, wecontrasted the result of restoration image with ray inspection image.The results showed that the use of ultrasonic testing can detect the no fusiondefects in brazed weld joins. But because of the effects in ultrasonic reflection ofelectronic system noise and grain doundaries, it makes the edge of defecs blurred. Especially when the defect size is small, it is more difficult to detect. After weused the image restoration technique, separation of noise and source signal isrealized and we got the clear defects image in the no fusion area and the edge ofthe image. The image which we got with ICA restoration technique is near to theray inspection result.