Vision‐based inter‐vehicle distance estimation for driver alarm system

Abstract

Determining the position of the vehicles on road is an important information in the driver assistance system which is used to increase driver's safety. The authors propose a robust real-time vehicle detection and distance estimation method for vision-based driving assistance system. They detect the vehicles in front of their own and estimate their three-dimensional position by using a single monochrome camera. After detecting the preceding vehicles using the common method, they use a new method for distance measurement based on monocular camera to estimate the inter-vehicle distance. The method also provides the longitudinal distance information for the driver alarm system. Finally, they utilise the Kalman framework to update the information of the vehicle which they detected before, the information includes horizontal distance, vertical distance and speed, and use these information to predict risk. Experimental results indicate that the system correctly detects vehicles in real time and effectively estimates the position of preceding vehicles.