STUDY ON VISION BASED PRECEDING VEHICLE DETECTION METHOD AT NIGHT

摘要

Preceding vehicle detection is a major research field of Safety Assistant Driving System. The paper proposed a preceding vehicle detection method based on multi-characteristics fusion of vehicle taillight at night. First, order-preserving method is selected to segment the images. Then the preceding vehicles are identified through the fusion of circular characteristics, symmetrical characteristics, edge characteristics and NMI characteristics of taillight. Experimental results have confirmed the reliability and effectiveness of the proposed method.