Conditional Variance of Differences: A Robust Similarity Measure for Matching and Registration
This paper presents a new similarity measure, the sum of conditional variance of differences (SCVD), designed to be insensitive to highly non-linear intensity transformations such as the ones occurring in multi-modal image registration and tracking. It improves on another recently introduced statistical measure, the sum of conditional variances (SCV), which has been reported to outperform comparable information theoretic similarity measures such as mutual information (MI) and cross-cumulative residual entropy (CCRE). We also propose two additional extensions that further increase the robustness of SCV(D) by relaxing the quantisation process and making it symmetric. We demonstrate the benefits of SCVD and improvements on image matching and registration through experiments.
KeywordsSimilarity Measure Mutual Information Reference Image Image Registration Conditional Variance
- 1.Irani, M., Anandan, P.: Robust multi-sensor image alignment. In: ICCV, pp. 959–966 (1998)Google Scholar
- 5.Pickering, M.R., Muhit, A.A., Scarvell, J.M., Smith, P.N.: A new multi-modal similarity measure for fast gradient-based 2d-3d image registration. In: Int. Conf. of IEEE Engineering in Medicine and Biology Society, pp. 5821–5824 (2009)Google Scholar
- 7.Pickering, M.R.: A new similarity measure for multi-modal image registration. In: International Conference of Image Processing, pp. 2273–2276 (2011)Google Scholar
- 8.Richa, R., Sznitman, R., Taylor, R.H., Hager, G.D.: Visual tracking using the sum of conditional variance. In: IROS, pp. 2953–2958 (2011)Google Scholar
- 9.Roche, A., Malandain, G., Pennec, X., Ayache, N.: The Correlation Ratio as a New Similarity Measure for Multimodal Image Registration. In: Wells, W.M., Colchester, A.C.F., Delp, S.L. (eds.) MICCAI 1998. LNCS, vol. 1496, pp. 1115–1124. Springer, Heidelberg (1998)Google Scholar