基于R分量的交通标志ROI提取

Traffic Sign ROI Extraction Based on Red Component

交通标志识别系统不仅能够利用前置摄像头帮助司机识别道路前方的交通标志,达到规范交通,减少交通事故的目的,在自动驾驶领域也有举足轻重的地位。不准确的交通标志感兴趣区域(region of interest,ROI)提取会极大地限制交通标志识别算法的准确率和实时性。在综合考虑了交通标志图像的RGB种基于红色分量(red component,R分量)的交通标志ROI提取算法,该算法将原始图像转移到RGBN空间(Normalized RGB),利用图像的红色特征和灰度特征相减后得到的图像的纯色信息提取ROI,和经典的基于HSV颜色特征的ROI提取算法相比,所提算法不仅克服了光照和天气变化对交通标志检测的影响,而且对最广泛的含有红色特征的交通标志ROI提取正确率提高到98.3%,有效比从10.3%提高到38.3%,实时性大幅度提高,为交通标志的高精度识别奠定了基础。

Traffic sign recognition system can not only help drivers identify traffic signs in front of the road with front-facing camera,achieve the purpose of standardizing traffic,reducing traffic accidents,but also plays a decisive role in the field of automatic driving.Inaccurate region of interest(ROI)extraction of traffic signs will greatly limit the accuracy and real-time performance of traffic sign recognition algorithm.Based on the comprehensive consideration of RGB color features and gray features of traffic sign images,this paper proposes a new algorithm for traffic sign ROI extraction based on red component(R component).The algorithm transfers the original image to Normalized RGB space and extracts ROI from the pure color information obtained by subtracting the red and gray features of the image.Compared with the classical HSV color feature based on ROI extraction algorithm,the proposed algorithm not only overcomes the influence of illumination and weather changes on traffic sign detection,but also improves the accuracy of the most widely used traffic sign ROI extraction with red features to 98.3%,the effective ratio from 10.3%to 38.3%,and the real-time performance is greatly improved,which lays a foundation for high-precision identification of traffic signs.