Image-based automatic traffic lights detection system for autonomous cars: a review

Sarita Gautam,Anuj Kumar

DOI: https://doi.org/10.1007/s11042-023-14340-1

IF: 2.577

2023-01-23

Multimedia Tools and Applications

Abstract:From the early stages of autonomous vehicle's development, traffic light detection/perception system have been an important area of research for making collision safe self-driving vehicles. Here Automatic Traffic Light Detection System (ALTDS) helps in accurate detection of Traffic Lights for Autonomous vehicles and Driver assistance systems (DAS). These vision-based system captures images using a camera mounted on a car and no other sensors. As traffic light is a small object in a real-time traffic scenario, so high-quality images are the main success factor of ATLDS. This paper elucidates the ideas and challenges that needs to be worked upon for better traffic light detection system used in self-driving cars. In this paper, we present a state-of-art review of various techniques used in traffic light detection. Different ATLDS techniques such as preprocessing, segmentation, feature extraction, classification and post-processing are categorized based on the features used at each stage, a comparison of the pros and cons of each technique is also provided. The hardware/software limitations, laws related to self-driving vehicles along-with simulation environments are also provided. The conclusion and future scope are given at the end.

computer science, information systems, theory & methods,engineering, electrical & electronic, software engineering