A Iso-Geodesic Stripes based similarity measure method for 3D face

How to evaluate the result of face reconstruction has always been a difficult issue. The similarity evaluation between original face and reconstructed face can be used to check reconstruction quality and explore new theories. The article puts forwards a similarity measure method which is based on Iso-Geodesic Stripes. First, overlap nose tip to eliminate translation difference and apply the classical PCA and ICP alignment algorithm to eliminate rotation difference. Then, simplify each 3D face with a series of Iso-Geodesic stripes, and each pair of stripes can be described by a distribution vector of 12 dimensions, which reflects 3D space distribution feature of two sets of vertices. Finally, the entire face is represented by a distribution matrix which consists of all distribution vectors. Therefore the similarity between two distribution matrixes indicates the similarity between two faces. The experimental results show that the similarity measure method is consistent with the subjective evaluation. In addition, the test results on SHREC2008 database also show it is an efficient and expression invariant in 3D face recognition method.