Jacobians for Lebesgue registration for a range of similarity measures

Research output: Book/ReportReportResearch

In [Darkner and Sporring, 2011] was presented a framework based on locally orderless images and Lebesgue integration resulting in a fast algorithm for registration using normalized mutual information as dissimilarity measure. This report extends the algorithm to arbitrary complex similarity measures and supplies the full derivatives of a range of common dissimilarity measures as well as their obvious extensions.
Original languageEnglish
PublisherDepartment of Computer Science, University of Copenhagen
Number of pages8
Publication statusPublished - 2011
SeriesKoebenhavns Universitet. Datalogisk Institut. Rapport

ID: 45953557