We propose a method that detects vehicles from in-vehicle monocular camera images captured during nighttime driving. Detecting vehicles from their shape is difficult at night; however, many vehicle detection methods focusing on light have been proposed. We detect bright spots by appropriate binarization based on the characteristics of vehicle lights such as brightness and color. Also, as the detected bright spots include lights other than vehicles, we need to distinguish the vehicle lights from other bright spots. Therefore, the bright spots were distinguished using Random Forest, a multiclass classification machine-learning algorithm. The features of bright spots not associated with vehicles were effectively utilized in the vehicle detection in our proposed method. More precisely vehicle detection is performed by giving weights to the results of the Random Forest based on the features of vehicle bright spots and the features of bright spots not related to the vehicle. Our proposed method was applied to nighttime images and confirmed effectiveness.
Yuta SAKAGAWA
Shizuoka University
Kosuke NAKAJIMA
Shizuoka University
Gosuke OHASHI
Shizuoka University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Yuta SAKAGAWA, Kosuke NAKAJIMA, Gosuke OHASHI, "Vision Based Nighttime Vehicle Detection Using Adaptive Threshold and Multi-Class Classification" in IEICE TRANSACTIONS on Fundamentals,
vol. E102-A, no. 9, pp. 1235-1245, September 2019, doi: 10.1587/transfun.E102.A.1235.
Abstract: We propose a method that detects vehicles from in-vehicle monocular camera images captured during nighttime driving. Detecting vehicles from their shape is difficult at night; however, many vehicle detection methods focusing on light have been proposed. We detect bright spots by appropriate binarization based on the characteristics of vehicle lights such as brightness and color. Also, as the detected bright spots include lights other than vehicles, we need to distinguish the vehicle lights from other bright spots. Therefore, the bright spots were distinguished using Random Forest, a multiclass classification machine-learning algorithm. The features of bright spots not associated with vehicles were effectively utilized in the vehicle detection in our proposed method. More precisely vehicle detection is performed by giving weights to the results of the Random Forest based on the features of vehicle bright spots and the features of bright spots not related to the vehicle. Our proposed method was applied to nighttime images and confirmed effectiveness.
URL: https://global.ieice.org/en_transactions/fundamentals/10.1587/transfun.E102.A.1235/_p
Copy
@ARTICLE{e102-a_9_1235,
author={Yuta SAKAGAWA, Kosuke NAKAJIMA, Gosuke OHASHI, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Vision Based Nighttime Vehicle Detection Using Adaptive Threshold and Multi-Class Classification},
year={2019},
volume={E102-A},
number={9},
pages={1235-1245},
abstract={We propose a method that detects vehicles from in-vehicle monocular camera images captured during nighttime driving. Detecting vehicles from their shape is difficult at night; however, many vehicle detection methods focusing on light have been proposed. We detect bright spots by appropriate binarization based on the characteristics of vehicle lights such as brightness and color. Also, as the detected bright spots include lights other than vehicles, we need to distinguish the vehicle lights from other bright spots. Therefore, the bright spots were distinguished using Random Forest, a multiclass classification machine-learning algorithm. The features of bright spots not associated with vehicles were effectively utilized in the vehicle detection in our proposed method. More precisely vehicle detection is performed by giving weights to the results of the Random Forest based on the features of vehicle bright spots and the features of bright spots not related to the vehicle. Our proposed method was applied to nighttime images and confirmed effectiveness.},
keywords={},
doi={10.1587/transfun.E102.A.1235},
ISSN={1745-1337},
month={September},}
Copy
TY - JOUR
TI - Vision Based Nighttime Vehicle Detection Using Adaptive Threshold and Multi-Class Classification
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 1235
EP - 1245
AU - Yuta SAKAGAWA
AU - Kosuke NAKAJIMA
AU - Gosuke OHASHI
PY - 2019
DO - 10.1587/transfun.E102.A.1235
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E102-A
IS - 9
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - September 2019
AB - We propose a method that detects vehicles from in-vehicle monocular camera images captured during nighttime driving. Detecting vehicles from their shape is difficult at night; however, many vehicle detection methods focusing on light have been proposed. We detect bright spots by appropriate binarization based on the characteristics of vehicle lights such as brightness and color. Also, as the detected bright spots include lights other than vehicles, we need to distinguish the vehicle lights from other bright spots. Therefore, the bright spots were distinguished using Random Forest, a multiclass classification machine-learning algorithm. The features of bright spots not associated with vehicles were effectively utilized in the vehicle detection in our proposed method. More precisely vehicle detection is performed by giving weights to the results of the Random Forest based on the features of vehicle bright spots and the features of bright spots not related to the vehicle. Our proposed method was applied to nighttime images and confirmed effectiveness.
ER -