Closed smeyel closed 10 years ago
Feature-based tracking RelWork: Kalman filtering, Kanade-Lucas-Tomasi Feature Tracker. Feature grouping: Nearest Neightbour, Normalized Cut (motion segmentation), common motion along trajectory lifetime. Current method: graph from feature point tracks as nodes, edges indicate common motion (max distance), connected components represent vehicle hypotheses!
N. Saunier ; T. Sayed The 3rd Canadian Conference on Computer and Robot Vision (CRV'06) 2006
AFeatureBasedTrackingAlgorithmForVehiclesInIntersections.pdf Feature point based tracking.