Detekce protijedoucích vozidel z obrazů za pomocí kamery

Abstract

This thesis is about car detection possibilities. The first part contains theory about the features used in object detection. Haar-like Features with modified version, LBP Features with improved MB-LBP and finally Strip Features are presented. Next, the classificator and boosting methods are described. The last part of the theory is about sliding window. Second part of this thesis is about creating operational object detector. OpenCV library is presented. The process of collecting images for training cascade classifier is described and training of course. Then these classifiers are used for testing on images and results are shown.