Automatic ROI Extraction in Casting Flaw Detection

摘要

Using a LoG operator to extract edges in casting flaw detection will produce many noise edges at the same time, which greatly affects the speed and precision of subsequent classification. In order to solve this problem, a method based on the maximum continuous subsequence algorithm has been developed as the preprocessing of casting flaw detection. Experiment shows that the method can effectively eliminate noise edges and preserve flaw contours as the KOI, which greatly reduces the subsequent processing.