Detection and Tracking of Point Features-英文文献.pdf

Shap e and Motion from Image Streams a Factorization Metho dPart Detection and Tracking of Point Features Carlo Tomasi Takeo Kanade April CMUCS Scho ol of Computer Science Carnegie Mellon University Pittsburgh PA This research was sp onsored by the Avionics Lab oratory Wright Research and Devel opment Center Aeronautical Systems Division AFSC US Air Force WrightPatterson AFB Ohio under Contract FC ARPA Order No The views and conclusions contained in this do cument are those of the authors and should not b e interpreted as representing the ocial p olicies either expressed or implied of the US government Keywords computer vision motion shap e timevarying imagery Abstract The factorization metho d describ ed in this series of rep orts requires an al gorithm to track the motion of features in an image stream Given the small interframe displacement made p ossible by the factorization approach the b est tracking metho d turns out to b e the one prop osed by Lucas and Kanade in The metho d denes the measure of match b etween xedsize feature windows in the past and current frame as the sum of squared intensity dierences over the windows The displacement is then dened as the one that minimizes this sum For small motions a linearization of the image intensities leads to a NewtonRaphson style minimization In this rep ort after rederiving the metho d in a physically intuitive way we answer the crucial question of how to cho ose the feature windows that are b est suited for tracking Our selection criterio