An Automatic Texture Mapping Method Using Local Surface Consistency Constraint

Due to the limitations of the accuracies in dense image matching and the factors of simplification during the three dimensional mesh generation, enormous noises normally occur on the triangular mesh surfaces during object reconstruction procedures. Furthermore, the geometrical and radiant differences between images, together with the noises, cause the textured mesh fragmentation. Aiming to solve this problem, this paper proposes a texture mapping method using a local surface consistency constraint. In this mapping procedure, fragmented triangles are merged into a relative larger surface with region growing, considering continuity and planarity. The regions are treated as a rigid surface and mapped to the same image to relieve the fragmented effect in the textured mesh. Experimental results show that the proposed method is effective and the output has better consistency than industry standard software such as street factory.