Development of a Multimedia-Based Vehicle Lane Departure Warning, Forward Collision Warning and Event Video Recorder Systems

Abstract

This paper presents how an intelligent multimediabased vehicle warning device is developed. The focal device incorporates the lane departure warning (LDW) function, the forward collision warning (FCW) function, and the event vide-recorder (EVR) function with a charged-coupled-diode (CCD) camera as the means to capture image. The LDW component uses a median filter, an edge-enhancement filter and the Hough Transform algorithm for lane recognition. The FCW component identifies vehicles with a featurebased approach while verifies the vehicle candidates by the appearance-based approach. Besides, we also propose a noble vehicle detecting scheme in which the task of vehicle detection depends not on the whole image within a frame but rather on the image's three constituent portions of different sizes, with a view to reduce the computing burden. The motion vector (MV) estimation is applied to track the detected vehicle in movement. This act helps that not all vehicles inside the image frame subject to detect in the vehicle detection stage. The EVR system is used to record the image captured in the event of a vehicle accident. The integration of LDW, FCW and EVR functions has successfully implemented in an ADI-BF561 600MHz dual core digital signal process (DSP). Ninth IEEE International Symposium on Multimedia 2007 -Workshops 0-7695-3084-2/07 $25.00