Microcrack detection method for plate heat exchanger based on machine vision

ZHANG Xin-peng1, WEI Li-feng1, YUAN Wei-qi2, DING Zhi-bo2
(1.College of Infromation Engineering, Shenyang University of Chemical Technology, Shenyang 110142, China;
2.Computer Vision Group, Shenyang University of Technology, Shenyang 110870, China)

Abstract: Aiming at the problems of high detection error rate and low realtime penetrate detection of microcracks in plate heat exchanger plates, the image acquisition, microcrack morphology, region segmentation, edge extraction and feature quantization were studied. A microcrack detection method was based on machine vision. The region of interest where microcracks may exist was segmented by the threshold algorithm. The Canny operator was used to extract the microcracks in the region of interest, and the spatial position features and local gray features of each rough extraction result were calculated. According to the defined features, the constraints were used to screen the results of each rough extraction to obtain true microcracks. The detection method was tested by the image collected by the factory. The results indicate that the method can meet the industrial requirements of noleakage detection and low false detection, and has realtime performance. It has certain practical application value for the automatic detection of microcracked on the plate heat exchanger plate.
Key words: machine vision; heat exchanger plate; microcrack; defect detection; edge detection; feature extraction