Efficient NCC-Based Image Matching in Walsh-Hadamard Domain
In this paper, we proposed a fast image matching algorithm based on the normalized cross correlation (NCC) by applying the winner-update strategy on the Walsh-Hadamard transform. Walsh-Hadamard transform is an orthogonal transformation that is easy to compute and has nice energy packing capability. Based on the Cauchy-Schwarz inequality, we derive a novel upper bound for the cross-correlation of image matching in the Walsh-Hadamard domain. Applying this upper bound with the winner update search strategy can skip unnecessary calculation, thus significantly reducing the computational burden of NCC-based pattern matching. Experimental results show the proposed algorithm is very efficient for NCC-based image matching under different lighting conditions and noise levels.
Keywordspattern matching image matching image alignment normalized cross correlation winner update
- 11.Di Stefano, L., Mattoccia, S.: A Sufficient Condition based on the Cauchy-Schwarz Inequality for Efficient Template Matching. In: IEEE International Conf. Image Processing, Barcelona, Spain, September 14-17 (2003)Google Scholar
- 12.Lewis, J.P.: "Fast template matching," Vision Interface, pp. 120–123 (1995)Google Scholar
- 17.Mahmood, A., Kahn, S.: Exploiting Inter-frame Correlation for Fast Video to Reference Image Alignment. In: Proc. 8th Asian Conference on Computer Vision (2007)Google Scholar
- 18.Pele, O., Werman, M.: Robust real time pattern matching using Bayesian sequential hypothesis testing. IEEE Trans. Pattern Analysis Machine Intelligence (to appear)Google Scholar