A flaw detection method based on morphological image processing

Mathematical morphology is used for image analysis for the purpose of flaw detection in materials that are represented by very poor-contrast digital images. In particular, an algorithm for flaw detection in the case of TV tube matte surfaces has been developed. The object skeletons within binary-valued images are obtained, and directional connectivity information in the skeletons is used to discriminate noise patterns from flaws according to a specified criteria. After the discriminating process, the skeletons that remain correspond to flaws and can be used to recover the actual shape of the flaws. An alarm flag may arise if the sizes of the detected flaws are found to exceed industrial standards. In the case of a gray-scale image, the image is first converted in a binary version using an adaptive threshold algorithm; then the flaw detection algorithm for binary images is applied. Experimental results are obtained for both binary and gray-scale digital image data obtained from imperfect glass samples.<
>