3D Traffic Sign Tracking Using a Particle Filter

Abstract

In recent years, there was much activity in the development of camera based active safety systems to aid and to support the driver of a car. One application for such a system is the detection and classification of traffic signs. An important aspect of such a system is the tracking of traffic signs. We present a novel algorithm to track traffic signs in 3D using a single monochrome camera. The algorithm allows to use the constraint that the observed movement on the image plane is entirely caused by the host car movement, which is partially known from internal sensors. The usage of the sensor information improves the tracking process and allows a robust rejection of false positive detections. We also present a way to incorporate a shape cue directly from the image plane into the tracking process. First tests show good results in practice and indicate, that this kind of tracking makes a very valuable addition to a traffic sign detection system.

Kernel-based traffic sign tracking to improve highway workzone recognition for reliable autonomous driving

“…For example, several researchers manually investigated color values, while others, including us, used machine learning techniques to obtain the optimal thresholds of target color [13]. The shape-based sign detections usually utilize either the gradient of gray scale image [14] or trained model from data sets [1].…”

“…The majority of the tracking systems deal with the dynamics of the tracked signs while the others cope with the appearances of the tracked signs. In order to reduce the search region and improve the detection performance, the discrete-time dynamics of vehicles have been utilized, for example, in Bayesian Filters [8], [11], [13], [14], [17] and information fusion [1], in order to predict the position of a sign in subsequent frames.…”

Kernel-based tracking for improving sign detection performance

“…Several researchers in the past elected to manually find the target color ranges [5], [9], which is simple but prone to error, while the others use machine learning techniques to obtain the optimal ranges of target color [10]. The shapebased sign detections usually utilize either the gradient of gray scale image [11] or trained model from sign shape data sets [1].…”

“…Among those systems, we divide them into two different approaches: one deals with the dynamics of the tracked signs, and the other copes with the appearance of the tracked signs. The dynamics approach for tracking is to utilize the discrete-time dynamics of vehicles, such as with Bayesian Filters [6], [8], [10], [11], [14] and information fusion [1] to predict the position of a sign in subsequent frames based on image frames received previously. This can reduce the computational cost, while still only relying on the detection performances, if the update model is defined by only the detection algorithm.…”

A framework of traffic lights detection, tracking and recognition based on motion models

“…Piccioli et al [9] and Fang et al [10] provide a Kalman-style tracker in the camera 3D coordinate system, but their works *The work was partly supported by National Natural Science Foundation of China ( are developed under the assumption that the vehicle moves absolutely straight. Fang [10] and Meuter et al [11] takes the rotation into consideration but without the consideration of yawing and pitching. Recently, some works take the advantages of the priori knowledge in order to track the traffic lights or traffic sign more efficiently, such as Schlosser et al [12] and Levinson [13].…”