A New Sequential Skeletonization Algorithm for 2-D Gray Images in Computer Vision Applications
Srinivasa Rao Perumalla, Yedukondalu Kamatham · 2018
Objects shape analysis, tracking, retrieval, matching, recognition, registration and compression, skeletonization technique provides a compact representation. It enables efficient assessment of local object properties such as scale, orientation, topology and so on. This paper proposes a new skeletonization algorithm for gray scale 2-D images. The proposed algorithm is based on repeatedly conditionally eroding the gray pixel values in the image until a one pixel thick skeleton is obtained. Erosion conditions are framed to preserve the connectivity. Initial experiments show that the proposed algorithm produces a comparatively good quality of skeletons for different shapes which are encountered in many real time applications such as computer vision. Results of applying this algorithm for a variety of images are encouraging.