Multi-modal Registration Using a CombinedSimilarity Measure

摘要

In this paper we compare similarity measures used for multi-modal registration, andsuggest an approach that combines those measures in a way that the registration parameters areweighted according to the strength of each measure. The measures used are: (1) cross correla-tion normalized, (2) correlation coefficient, (3) correlation coefficient normalized, (4) the Bhat-tacharyya coefficient, and (5) the mutual information index. The approach is tested on fruit treeregistration using multiple sensors (RGB and infra-red). The combination method finds theoptimal transformation parameters for each new pair of images to be registered. The methoduses a convex linear combination of weighted similarity measures in its objective function. Inthe future, we plan to use this methodology for an on-tree fruit recognition system in the scopeof robotic fruit picking.