Abstract: The initial model computation is a crucial step in image sequence based 3D surface reconstruction using deformable model,this paper presents a visual hull method for computing the model from multi-images.On the basis of snake deformable model,the algorithm translates the visual hull computation problem into a problem on force computation.Firstly,the silhouettes are extracted from multi-images,and the internal and silhouette forces are computed from the smoothness and the contours information.Then the initial mesh is deformed under the internal and silhouette forces until convergence from a sphere to the visual hull.In the process of deformation,the mesh optimization methods,such as edge collapse,split and flip,are used to prevent from mesh confusion.Our deformable approach reduces the mesh mistake,makes it easy to implement,parameters tunable,and memory cost efficient.Experimental results show that the final mesh has higher precision and smoothness than the traditional schemes.