A Comparative Study of Vision-Based Traffic Signs Recognition Methods
Traffic signs recognition is an important component in driver assistance systems as it helps driving under safety regulations. The aim of this work is to propose a vision based traffic sign recognition. In the recognition process, we detect the potential
- PDF / 2,503,770 Bytes
- 8 Pages / 439.37 x 666.14 pts Page_size
- 56 Downloads / 188 Views
)
MIRACL-FS, Sfax University, Rte Sokra Km 3, BP 802 3018 Sfax, Tunisia [email protected], [email protected], [email protected] 2 National Engineering School of Gabes, Gabes, Tunisia [email protected]
Abstract. Traffic signs recognition is an important component in driver assis‐ tance systems as it helps driving under safety regulations. The aim of this work is to propose a vision based traffic sign recognition. In the recognition process, we detect the potential traffic signs regions using monocular color based segmen‐ tation. Afterwards, we identify the traffic sign class using its HoG features and the SVM classifier. As shown experimentally, compared to leading methods from the literature under complex conditions, our method has a higher efficiency. Keywords: Traffic sign detection · Traffic sign recognition · SVM classifier
1
Introduction
Automatic traffic sign recognition (TSR) is an important task for an Advanced Driver Assistance System (ADAS). In fact, some higher-end car models already offer such system [1]. The traffic signs colors and pictogram make them easily perceived and interpreted. Therefore, they raise driving safety by warning against danger and difficul‐ ties around the drivers and help them with their navigation by providing useful infor‐ mation. However, recognizing traffic signs in out-door context is still a challenging problem as it has to overcome weather conditions, traffic sign appearance variation and real-time processing constraint. Actually, the traffic sign recognition performs on two steps: detection and classifi‐ cation of Traffic Sign Detection. In the detection step, the image is segmented relying on the visual key of traffic signs features such as color [2, 3] and shape [4]. Once the candidate traffic sign regions have been detected, a classifying step is performed to make the decision to keep or reject a candidate region of traffic sign. To ensure a prominent classification, there are training-based methods [5, 6] and model-based methods [7]. In our work, we have defined the appropriate methods to use in the proposed solution for traffic sign detection and classification. For the detection step, we opted for a color based method since it provides a faster focusing on the potential areas of traffic signs. In fact, similar objects to the traffic signs shapes may coexist in the background like windows, mail boxes and cars. Besides, methods based on shapes require robust edge detection algorithm which is not an easy task with a not head-on viewing angle or with © Springer International Publishing Switzerland 2016 A. Campilho and F. Karray (Eds.): ICIAR 2016, LNCS 9730, pp. 341–348, 2016. DOI: 10.1007/978-3-319-41501-7_39
342
N.B. Romdhane et al.
low resolution traffic sign capture. For classification step, we used an SVM classifier thanks to its performance in statistical learning theory and robustness already proved in TSR topic. The remainder of this paper is organized as follows: Sect. 2 presents the proposed method. Section 3 discusses a set of extensive experimental evaluations and compares the
Data Loading...
