Image Matching Algorithm Based on Subdivision Wavelet and Local Projection Entropy

An image matching algorithm that uses subdivision wavelet and the local projection entropy of image is proposed and tested in this paper. Subdivision wavelet, which is of the second generation wavelet, has much lower computing cost than the first generation wavelet due to the nonentity of the shifts and dilations. The initial properties of signal can be retained when signal is transformed by subdivision wavelet. The pixels of image can be doubly reduced when image is decomposed by subdivision wavelet. The local projection entropy of image, which expresses the local features of image, not only has the ability of resisting noise but also can depress the feature dimension of image, so the computational cost of image match can be reduced. In the algorithm the real time image and template image are decomposed respectively through subdivision wavelet. Then the match is performed between the decomposed real time image and the decomposed template image using the local projection entropy of image. Experiment results show that the algorithm can effectively promote the image matching speed and possesses the stronger ability of resisting noise under retaining a certain precision