제목CHANNEL FEATURE BASED VEHICLE DETECTOR THROUGH
VEHICLE REGION ESTIMATION USING HAAR-LIKE FEATURE
저자명

SANGHUN HAN, DUYEOL YU, HERNSOO HAHN, YOUNGJOON HAN

Abstract This paper proposes a vehicle detector that finds a candidate position of the forward vehicle and verifies the vehicle candidate using a channel feature. Detection of a vehicle through learning searches the entire image and thus increases detection time. Therefore, we select a candidate position to restrict the search region and detect vehicles based on the channel feature. To select candidates, the region of interest is determined through horizontal and vertical edge projection,and the region is clarified by shadow and symmetry. The clarified candidates are verified by the Adaboost algorithm based on the channel features. The channel feature in this method of vehicle detection is robust because the LUV channel in the feature is prominent at the tail lamp. To evaluate the performance of this algorithm, we experimented on image sequences in various driving environments.
원문 수록처

Proceedings of ISER 59th International Conference, Prague, Czech Republic, 21st -22nd June 2017

p.10-12

자료유형International Conference