Project for automatic traffic light signal recognition. Based on OpenCV (here is used 2.4.4), C++ interface. Traffic light position is assumed to be static, but can be re-defined manually at runtime.
For distortion compensation, Lucas-Kanade algorithm is proposed.
Useful links and resources:
RUS. Haar cascades, Viola Jones, OpenCV http://habrahabr.ru/post/67937/ http://habrahabr.ru/post/133826/ http://habrahabr.ru/post/208092/
RUS. VS2010 plus Qt plus OpenCV (will be used for thesis project demo) http://www.troyashka.ru/2011/05/qtvs2010opencv.html
ENG. Vehicle detection projects of a cool guy http://www.behance.net/gallery/Vehicle-Detection-Tracking-and-Counting/4057777
ENG. Big research on car detection (with learned cascades! and videos) http://lars.mec.ua.pt/public/Media/ http://lars.mec.ua.pt/public/Media/HaarFeatures_RoadFilms/HaarFeaturesTests/ http://lars.mec.ua.pt/public/Media/HaarFeatures_RoadFilms/