Pruning redundant skeleton branches of object in image

Skeleton is one of the most important features in image processing. In many applications such as matching, animation, tracking and so on, finding main features are important; so, obtaining target skeleton can extract suitable target features. In this paper we try to introduce a fast and accurate algorithm to achieve main skeleton of each objects. Therefore, we suggest an appropriate approach for pulling out proper skeleton. We claim our algorithm stability is strong enough to confront edge noises. Our proposed method based on Contour Length Measure. At First, we extract object´s skeleton software that we called it M-Skeleton. Second, redundant branches would be pruned by checking relevance rate for all edge pixels of target picture. So the remained branches make target´s main skeleton. Most of presented skeleton algorithms are dependent on adjusting threshold, but our proposed algorithm is almost independent and experiments exhibit it can truly extract target body skeleton.