Inter-vehicle distance detection based on keypoint matching for stereo images

Abstract

An algorithm to detect car distance from a pair of stereo images is presented. It is useful for drivers to avoid collisions and ensure safety to keep the car at a constant distance from the car ahead. The conventional distance detection method is based on image matching; the proposed algorithm is based on key-point matching. Key points are extracted from a stereo image pair by using Speeded Up Robust Features (SURF). The distance is calculated from 3D binocular disparity, the difference of position at the object.