TY - GEN
T1 - Öznitelik betimleyicileri füzyonu ile trafik işaretlerinin tespit edilmesi ve taninmasi
AU - Erhan, Can
AU - Tazehkandi, Amin Ahmadi
AU - Yalçin, Hülya
AU - Bayram, Ilker
PY - 2013
Y1 - 2013
N2 - This paper presents an algorithm for most prominent component of active vehicle safety applications, namely the detection and recognition of traffic signs. In the detection stage, HOG feature descriptors combined with SVM classifiers are used to determine the location of points that are high likely to be the potential traffic signs in the scene. Once the search space for traffic sign recognition is reduced through first stage, SURF, FAST and Harris algorithms are used to extract the keypoints in these potential traffic sign regions and BRIEF feature descriptors are used to define the neighbourhood around these keypoints. Model traffic signs are then compared to the regions that are detected to be potential traffic signs in the current traffic scene to determine the type of the traffic sign. In order to extract keypoints, the performance of a variety of feature descriptors are analyzed. Proposed method is tested on video sequences acquired by the camera mounted on a vehicle cruising inner city traffic.With %90 success rate, experimental results suggest that SURF algorithm outperforms the other algorithms in recognizing traffic signs.
AB - This paper presents an algorithm for most prominent component of active vehicle safety applications, namely the detection and recognition of traffic signs. In the detection stage, HOG feature descriptors combined with SVM classifiers are used to determine the location of points that are high likely to be the potential traffic signs in the scene. Once the search space for traffic sign recognition is reduced through first stage, SURF, FAST and Harris algorithms are used to extract the keypoints in these potential traffic sign regions and BRIEF feature descriptors are used to define the neighbourhood around these keypoints. Model traffic signs are then compared to the regions that are detected to be potential traffic signs in the current traffic scene to determine the type of the traffic sign. In order to extract keypoints, the performance of a variety of feature descriptors are analyzed. Proposed method is tested on video sequences acquired by the camera mounted on a vehicle cruising inner city traffic.With %90 success rate, experimental results suggest that SURF algorithm outperforms the other algorithms in recognizing traffic signs.
KW - Active vehicle safety
KW - Traffic sign recognition
UR - http://www.scopus.com/inward/record.url?scp=84880900132&partnerID=8YFLogxK
U2 - 10.1109/SIU.2013.6531427
DO - 10.1109/SIU.2013.6531427
M3 - Konferans katkısı
AN - SCOPUS:84880900132
SN - 9781467355629
T3 - 2013 21st Signal Processing and Communications Applications Conference, SIU 2013
BT - 2013 21st Signal Processing and Communications Applications Conference, SIU 2013
T2 - 2013 21st Signal Processing and Communications Applications Conference, SIU 2013
Y2 - 24 April 2013 through 26 April 2013
ER -