The Turns Light Recognize Based Vision Survey

With the fast development of economy, our country steps into the automobile society, traffic safety and unobstructed becoming questions that societies generally pay attention to. In this case, people begin to put into the intellectual traffic system (ITS). while studying, the intellectual traffic system (ITS) includes the intellectual infrastructure and intellectual vehicle, its key technology involves pattern-recognition, pattern process, digital signal process, artificial intelligence , electronic technology, information technology, communication technology and system engineering .etc., it is a comprehensive discipline.The environmental recognition system is a important component of the intellectual vehicle, it gathers and recognize various kinds of vision goals appearing while the vehicle goes, can extract the front state in time, point out or warn to the driver, or directly control the vehicle, in order to keep the traffic unobstructed and prevent the emergence of the accident. In various kinds of environmental parameters, various kinds of automobile signal lights of the front vehicle are undoubtedly very important. In these signal lights, the turn light is one of the most important signal light in the intellectual car when it turns, change road, etc. The research work that the light recognizes is significant.At present, the research work that the turn light recognizes was launched less both at home and abroad, such a system has not still come out yet either. But, the work in this respect has already been launched because the light recognize is very importance inherent advantage of the system based on vision to turn, We can predict, the work in this respect will be paid close attention to by researchers of various countries soon. In turn light recognize, there still are lots of questions need to deepen systematic research.The main content that this text studies includes: Produce and upgrade algorithms based background of statistic model; The extract and express of vehicle based two value picture; The split shadow algorithm based two value; The light split algorithm based Laplacian operator and HSI space; The turn light extract based balance analyze; Light class algorithm. In producing and upgrading algorithms based background, improve the newer model of the background, raise the renewal speed of the background , guarantee the validity of the background at the same time ; In vehicle extract, propose merge method and gain a shadow remove algorithm Dealing with the two value picture, because this algorithm is based two value picture, it is very fast in speed; In turn light recognition, propose Laplacian operator, use this operator to recognize turn light based HSI color space, put forward and use the balance analyze method to extract the turn light. At the end of research work, based the position characteristic and moving characteristic, we get the type of turn light.