A recognition approach of 3-D objects based on the Tsallis entropy

As an attractive descriptor of 3-D object surfaces, spin image has attracted a considerable amount of researchers´ attention. But there are several unresolved problems about the spin images, such as the sharp increase of data amount. By using the Tsallis entropy (TESI), a concise representation of spin image is proposed. This representation can convert a spin image to a scalar by using the Tsallis entropy of the spin image, so that a 3D object can be described by using a vector which is composed of the Tsallis entropies. Furthermore, the spin images from new algorithm are rotation invariant, translation invariant and scale invariant in correlation sense. We discussed the application of TESI, and the experiments results can prove the efficient of the approach.