Morphological decomposition of arbitrarily shaped images

Decomposition of images is a very important issue in pattern analysis and recognition. Especially, for image processing systems that can not handle large size of images, image decomposition is the only way to overcome this difficulty. The technique presented in this paper is based on mathematical morphology and decomposes a binary structuring element into dilations of smaller size of images (factors). Park and Chin (1995) proposed a method of morphological decomposition of simply connected images into 3×3 size factors. We extend their theorem to make it possible for any n×n size factors. A method for optimal decomposition is also discussed. Based on this method, users can therefore select their favorable shape of structuring elements to process images efficiently