| Fluorescent magnetic particle inspection is a kind of conventional non-destructive testing technology, which has been widely used in many fields. Automation and intellectualization are two directions of the modern non-destructive testing technology development. In particular, automation and intelligent recognition are especially needed by the fluorescent magnetic particle inspection. In this paper, the image processing technology and pattern recognition technology are adopted in the fluorescent magnetic particle inspection, moreover, the author conducted the research with the combination of theory and practice.This paper uses image processing algorithm, principal component analysis(PCA) and Back Propagation(BP) neural network to design the fluorescent magnetic particle intelligent detection system and sets up a platform image acquisition of fluorescent magnetic particle. The techniques of image denoising algorithm and threshold segmentation algorithm for image preprocessing are firstly studied in this paper. Adaptive center weighted median filter algorithm and local threshold segmentation algorithm are selected for image preprocessing. By image enhancement is also studied the morphological operation technology. Finally principal component analysis for image feature extraction and the back propagation algorithm of Back Propagation neural network classifier design are both applied in this paper, with the purpose of completing the automatic defect recognition and classification.By analyzing the experimental results, it shows that the proposed algorithm can effectively recognize and classify the zero defect, the crack defects, bubble defects, the pseudo defects and slag defects. The overall recognition rate of the above five types has reached the requirement of 85%, particularly the recognition effects of the zero defect, the crack defects and bubble defects have reach 90%, and the overall omission rate has reached the requirement of below 5%, which presents certain practical value. |
PDF Full Text Request