Vision System Of Out-Door Vehicle For Guide-Line And Obstacle Inspection

The navigation of mobile vehicle has two main tasks to keep the vehicle runningon the road and avoiding the obstacle in front of the vehicle. This thesis introduced anew method of road and obstacle detection based on stereoscopic vision;it can testthe road region and obstacle precisely and immediately.Firstly, with the idea of building an open testing system, we adopts the "PC +adapter" structure which uses DMC (Digital Motion Controller) and IPC as thesupport unit of the system.The software is developed by using the OOP programming method, and themultithread is adopted in software design, this paper use Visual C++6.0 as thedevelopment tool to finish the inspection control program of the vehicle.In order to make preparations for the calculation of the navigation vision, thecalibration is made for the finished vehicle.This paper adopts the arithmetic based on edge detection to find the edge of lanemarking, by using this method we can achieve the inerrant driving of the vehicle.The models of camera are established, the calibration is made, and the two imagepoints on different images which are from one object point are matched. We usestereoscopic vision to measure 3D accurately, and we can detect the obstacle of therunning road immediately which can make the vehicle run safely.