This paper presents a framework of surface modeling from multi-view range data. The input to the algorithms are triangle meshes, each of which is from a single view range scan. The triangle meshes generated from raw data are first processed by the proposed area decreasing flow for surface denoising. Although the proposed flow is mathematically equivalent to the mean curvature flow, it can avoid the difficulty in curvature estimation and provide an optimal flowing step size. We introduce an adaptive triangle mesh smoothing scheme based on crease edge strength of each vertex using tensor voting of the normal vector field inside a geodesic window. The smoothing result makes surface normal estimation more accurate which is then used in surface mesh integration. Based on Hilton's implicit surface-based method, surfaces from multiple views are integrated into a single 3D model. We incorporate color images to generate textured models. The algorithms are successfully applied to surface modeling from range data using two types of range scanners.