| In recent years, with the increase of the popularity of cars, pedestrian traffic safety and efficiency problem have become increasingly prominent. In order to explore an effective way to reduce traffic accidents, Crosswalk signals recognition based on machine vision technology has become the current hot research topic, besides, it also provide certain theoretical basis for the blind help line tool.This paper presents detection and recognition method of pedestrian crossing traffic signal lamp under complex scene, the method not only can identify the detection of pedestrian crossing traffic signal lamp under complex scene, but also the specific state of light. Specifically, based on the sufficient research, summarized on the basis of pedestrian crossing traffic signal lamp, attribute, firstly, a framework for signal location and recognition system was established, and the system framework is divided into three parts,including image acquisition, positioning, recognition algorithm, the ARM hardware implementation. Image acquisition is divided into image acquisition and compression, the frame structure is simple, easy to implement. Based on digital image processing of the Crosswalk signals recognition and localization algorithm is the focus of this article.In view of the lack of simple scenario in the existing traffic light detection method, this paper puts forward a method based on RGB and HSI color space with the combination of threshold segmentation algorithm to locate and identify Crosswalk signals, and the algorithm is implemented with Matlab programming. Most of the current traffic lights recognition algorithms can only be implemented on PC, and cannot be applied to embedded mobile devices, for this shortage, this paper proposed and implemented in C++program to realize the algorithm, and after cross compile successfully transplanted to ARM9development board which Carries liunix operating system,completing the hardware implementation of the algorithm. |
PDF Full Text Request