Design Of Pedestrain Detection Based On Vehicle Auxiliary Driving System

In recent years, with the increasing of vehicle ownership, public tr ansportsecurity is facing tremendous pressure, how to avoid traffic accidents occurfrequently become an important research topic,the auxiliary driving is aneffective way to solve the problem of traffic safety. The pedestrian detectionbased on on-board video as an important module of the auxiliary driving system,make up the defects that the ability of getting image information in the nightdecreasing.In this paper, the design of a vehicle auxiliary driving system based onautomotive vision sensor and A5s30A processing chip, the system featuresinclude nighttime pedestriandetection, lane departure detection, road signrecognition, and prompt early warning function.Taking advantage of Opencv, trained a12-level pedestrian detector based onAdaBoost algorithm and picture set of positive and negative samples. And thedetector is used to realize the pedestrian detection function of the video image onthe core to A5s30A processor.On the basis of realizing the function of the pedestrian detection, optimizedthe pedestrian detection module. For detection rate is too low, using a variablestep size and partition scanning tracking methods to improve the detection rateand basically reached the requirements of real-time detection. for the false alarmrate is too high, used histogram equalization pre-processing andmulti-scaleintegration technology, histogram equalization effectively eliminates the effect offactors such as light and shadow effect, multi-scale integration technologycombined the detected rectangles, eliminate redundant portions of the testresults, and ultimately improvethe detection rate and reduce the false alarm rate.