Research Of Traffic Lane Change Video Detection Based On ARM

Convenience of vehicle brings much convenience for our life, but with deteriorating traffic condition, illegal lane change, overspeed and overload, drink driving and other acts cause many traffic accidents that could have been avoided. Therefore, more and more people propose that traffic system shall be maintained better. Video detection technology has features of high precision, low error and high efficiency, widely used in modern intelligent traffic network. Generally, relevant processing operation is made based on computer for most video detection technologies, which needs long –distance information transfer and large investment of fund; in addition, data information has delay and distortion during transfer. With rapid development of embedded system(ARM), by virtue of advantages of high stability, low cost and low power consumption, ARM products have become an integral part in life and work of people, so ARM based vehicle detection system has become the mainstream trend of intelligent traffic system(ITS).With the real –time traffic image shot by the camera installed on the road as major research object of exploration and analysis of this paper, make statistical analysis on the information about lane change of vehicles in video image, and conduct analysis and discussion on relevant object collection, shadow removal, information transfer and the like of foreground target. Besides, this paper also discusses the popular ARM system and ARM processing equipment at present, and among embedded systems of Linux, finally determines to adopt ARM11 micro processor for achieving operation of the above algorithm, design a set of system structure easy for installation and maintenance. Contents of this paper include:1.In terms of detection of driving vehicles, compare the advantages and disadvantages of optical flow method, inter –frame difference method and background difference method through practices, and research the algorithm through selecting image background attraction algorithm based on background difference as foreground vehicle object of system.2.In terms of shadow removal of foreground driving vehicles, describe the cause of appearance of shadow, and demonstrate the effect of shadow on foreground object through a series of relevant experiments. Under HSV color model space, deploy relevant experiment detection analysis for the shadow part, it is able to know that this detection method is efficient and accurate, meeting design requirements of system. The quality of image can be affected in shadow removal stage, so this paper also discusses morphological analysis on and solution to binaryzation image. In the detection process of lane change of vehicles, the key is conducting contrastive analysis on virtual detection line method and zone marking method, and finally the virtual detection line method with excellent precision and strong timeliness is selected as the main detection means.3. Discuss and analyze embedded operation system, according to the design demands analysis summarized in this paper, select code open source embedded system Linux(ubuntu) as system for detection of lane change of vehicles, at the same time, S5PV210 ARM11 processor of Samsung is selected as the processing equipment of this system.Discuss programming tool elements and operating steps of ARM processor, and make design scheme usage procedure of ARM processor end, with ability of collection, processing, analysis, storage and display of image data.4. In actual road traffic, all –around detection is given to this system, and detection results show that this system has good timeliness, accurate statistical results, stable performance and precise data transfer, meeting expected standards.