A novel synchronization feature template scheme based on coordinates of feature point set

This paper presents a novel synchronization feature template scheme against similarity transformation. In the scheme, the template is a point set, which only contains the coordinates of the points. The scheme is based on a new point set registration method proposed in this paper. The method mainly adopts four techniques: rotation, scaling and translation (RST) invariant geometric descriptors; stratified matching; online clustering and parameters average, called GSCA. Compared with those existing synchronization feature templates which contain point descriptors, the template scheme is more efficient in template storage and transfer. Theoretical analysis and experimental results prove that the scheme not only has both strong robustness and high precision to synchronize, but also has lower computational complexity. The GSCA technique is also suitable for feature-based content authentication, image registration, etc.