Research On Main Target Recognition And Vehicle Distance Detection Algorithms In Adaptive Cruise

With the continuous increasing of the vehicles in China,the number of automobile traffic accidents is also increasing.Adaptive cruise is an important part of automobile safety technology.As the eyes of the advanced driver assistance system,the sensor played a very important role in it.The different sensor can capture different information.The traditional adaptive cruise system uses millimeters wave radar.It is usually aimed at some particularly typical road conditions for straight-line driving.In order to solve the problem of road information acquisition and distance detection for adaptive cruise in complex road conditions,In this paper,the camera is used as the carrier to obtain the front multi-target information and lane lines by the methods of deep learning and image processing,then the linear camera model is used to calculate the target distance,finally identified the main target recognition.This article mainly focuses on the following,(1)Use the camera to obtain the road environment information by the deep learning and Image Processing(2)Use linear camera model to obtain the information of lane line and target position.The target coordinates are obtained by solving the equations established by the coordinate transformation.(3)According to the estimation of the center point of the lower edge of the rear of the vehicle,confirm the distance detection point in the photo,then calculate the point in the geodetic coordinate system by the relationship between the coordinate systems,finally According calculate the distance of the target vehicle.(4)According to the research of the coordinate system conversion relationship established by the linear camera model,the horizontal and vertical ranging experiments under two different conditions of static and dynamic are designed.The experiment mainly compares the horizontal and vertical true distance of the test point in the geodetic coordinate system and the horizontal and vertical calculated distance of the linear camera model converted to the geodetic coordinate system in the static situation.The experiment mainly compares the distance errors measured by different precision sensors in the movement of the target in the dynamic situation.(5)Considering the influence of the change of the pitch angle on the ranging model under static and dynamic conditions,methods of online calibration and KALMAN filtering are respectively proposed to improve the accuracy of the ranging model.