Vision-based estimation of range and direction of preceding vehicle for advanced driver assistance systems

Abstract

A vision-based system for range and direction estimation of preceding vehicle is proposed. The proposed system-on-chip system exploits the property of 2-D sliding DFT for edge detection. The horizontal edge map is obtained through the 2-D SDFT edge detector and the edge map is further processed for the vehicle detection, range and direction estimation of preceding vehicle. The vehicle detection is performed with edge detection followed by morphological erosion process, and the range and direction estimation is determined using imaging geometry approach. The FPGA implementation of the proposed algorithms are carried out with real-time conditions. The performance of the proposed methods for indoor and outdoor environments have also been investigated. © 2019 IEEE.