A real-time rear view camera based obstacle detection

Abstract

This paper presents a robust real-time rear-view camera based object detection algorithm for backup aid and parking assist applications. The system is capable of handling the challenges of stationary as well as moving objects in rear view of the host vehicle, utilizing a single car-mounted rear-view fish-eye camera.A motion-based and edge-based object detection algorithm was developed in order to detect near distance objects and distant objects respectively. Furthermore, a free-space detection algorithm was developed which is used to detect areas where no objects appeared in front of the vehicle path.A tracker-based fusion algorithm was used to track and fuse detection results of individual detection algorithms.Experiments have been carried out by applying the proposed algorithm on prerecorded sequences as well as within a test vehicle and thus in a closed loop environment. The experimental results indicate promising detection performance.

An Indoor Obstacle Detection System Using Depth Information and Region Growth

“…There are many proposed methods for the first category, such as [ 3 , 4 , 5 , 6 , 7 , 8 ]. Ma et al [ 3 ] proposed an object detection algorithm that uses edges and motion. The motion-information is used to determine the dynamic obstacles and the edge-information is used to determine obstacles.…”

Monocular Rear-View Obstacle Detection Using Residual Flow

“…In [3] a system is presented that uses monocular input and egomotion from odometry sensors. The authors in [4] present a system for backup aid that detects obstacles from three independent methods, fused at the output stage. It assumes road boundaries are always within the video image and has some difficulty with false positives due to shadows not being totally eliminated.…”