Based on 3D regional growing method, a novel surface tracking algorithm is proposed. Firstly, the seeding cubes, representative to different edge surfaces, are selected. Then, the edge surfaces lost during the de-noising step are recovered based on the connectivity of edge surfaces. Based on dynamic link stack, the DFS, a time and space efficient traversal algorithm, is employed during the ... [Show full abstract] tracking process. The experimental results show that the proposed tracking algorithm can overcome the defect of the traditional edge detecting methods which usually produce holes in the extracted surface. In addition, the proposed algorithm is shown to be able to reconstruct more accurate deformable surface models of edge surface in comparison with the equivalent surface extraction algorithm.