Spatial smoothing based segmentation method for internal defect detection in X-ray images of casting components

Abstract

X-ray imaging based inspection is an established non-destructive method for automatic detection of internal defects such as blow-hole, cold fill, shrinkage and foreign object inclusions in aluminium castings. For online inspection of casting components in the production line, X-ray imaging system needs to be integrated with dedicated image processing methods especially developed for automatic flaw detection. The image profile variation of captured image causes difficulty in segmenting the internal defects. In order to overcome this gray scale morphological variation, a spatial smoothing based effective segmentation method for detecting blow-hole and cold fill was proposed. Different image segmentation methods like Otsu, adaptive threshold and median filter were applied to the X-ray image samples and their defect detection results were compared and discussed. Our experimental results indicate that the developed method possess 100% detection accuracy in analyzing the internal casting defects relatively when compared with other segmentation methods which gives accuracy less than 95%. The proposed method very well improves the identification and detection rate of internal defects found in casting components.