Design And Implementation Of Vehicle Safety System In CHCD

China now is the largest developing country in the word, and its automobile industryhas growth rapidly in the past ten years.. Along with the great convenient humans benefitfrom vehicle, the high frequency of traffic incidents and high mortality are also botheringus. People paid their attention more&more on road safety. The traditional securitymeasures such as bumper, seat belts, air bags and other passive measures can noteffectively avoid the generation of traffic accident, then the active safety products such asvehicle safety system because it can nip in the bud, it has attracted increasingly attentionfrom both car manufacturers and consumers.At present, many abroad brand cars have been standard equipped the active safetyproducts, while in China due to the automotive industry started relatively late, in vehiclesafety products filed especially active safety products we did not see any report. Thispaper is to design the active safety products independently, filling the blank in the field,meanwhile taking this as a breakthrough for enterprise strategy transformation. On thosebasis, enterprise to expand its develop in automotive electronics, truly achieve thediversification of vehicle safety products, to follow the pace of worldwide development ofautomotive industry.This system is an lane departure warning system and forward collision warningsystem which based on camera information capture system. It used TI6347DSP as theplatform core, finished full integration of embedded and algorithm software. In theembedded software, this system considered various conditions, modes of operation,designed it as many modules and achieve the interactive function among the modules. Inthe algorithm, used subregional gray, adaptive threshold Sobel edge enhancementmethod for image pre-processing, used hough transform and Kalman filter for lanedetection and tracking. To reduce the computing time, it used dynamic search region, and used TLC (time to lane crossing)method for warning. As for the vehicle, system detect itvia its tail characteristics and bottom shadow, used the method of AdaBoost trainingvehicle classifier, and the TTC(time to collision) method to realize forward collisionwarning. In the calculation of the distance, system used the pinhole imaging principle, viathe mutual relationship between the world coordinate, camera coordinate and the imagecoordinate to calculate the lane and vehicle location. And for camera calibration, thispaper described the principle of camera calibration, and developed a simple method forcalibration based on corner detection, which reduced the environment dependence andshortened the calibration time. Meanwhile, system developed the user interface for easeuse and operation, which can alarm display and finish setting up a variety of system.Finally used racelogic VBOX devices to test for alarm accuracy of products, and analyzedthe test results, which showed that this alarm system meet the standard requirements andhad high reliability.With the development of automotive technology, automotive satety electronicproducts will have larger development space.