Along with the development of image processing technology and artificial intelligence,also supplied with the processing power of embedded system,the Advanced Driver Assistant System is coming into the market,such as Lane Departure Warning,Auto Cruis Control,Intelligent Speed Limit Warning and Auto Emergency Break.Especially the detection of the object surround and in front of the ego vehicle,is the key technology.Smart camera with the environmental perception ability,as the key component,the improvement of its technology is essential for the Advanced Driver Assistant System and also autonomous driving implementation.But as the state of art,mono camera has the advantage in size,power consumption,cost and reliability than stereo camera,is widely used.But it is hard to detect the precise position of the object.Designed for Advanced Driving Assistance area,the solution in this thesis is based on sequential frames of pictures and ego motion information.It provides a solution of estimating 3D position of objects on road and their trajectory.Consider the automotive component manufacture and vehicle assembly procesdure,this thesis use a newly designed method to calibrate the intrinsic and extrinsic parameter of the camera.In the technology of road surface reconstruction and object boundary,position,trajectory estimation,the optimization technology for real time process of the embedded syste is based on the SOC adopted by the product.The test result shows that the performance is close to the stereo camera.Finally,the thesis conducts the tests based on 2 levels,module level and system level.Especially the estabilishment of D-GPS and relative positioning system,the test was implemented on real road and vehicle for both function and performance test.The final result shows that the proposed solution can estimate the 3D position which is equivalent to stereo camera on flat road.On inclining road,the performance is closed to stereo camera.This result shows that the soluation described in the paper is very useful according to the state of art. |