Abstract: Second selection method is presented for extracting contour dominant points of industrial CT image, which is simple and efficient. The method comprises rough extraction and accurate extraction. First discrete circular curvatures of contour points are calculated. Bad points are removed by rough extraction, meanwhile, points which aren’t correlated with contour feature are also removed. Then in the process of accurate extraction, contour dominant points are extracted by levels of detail. Experimental results show that contour dominant points can express contour’s shape and redundant data are removed.

Key words: Discrete curvature, Dominant points, Industrial CT image, Levels of detail analysis