自然背景下交通标志牌的检测方法研究  

Research on Traffic Sign Detection Method in Natural Background

作　　者：郑挥 郭唐仪[2] 王建博[3] 朱周[2] 

机构地区：[1]苏交科集团股份有限公司,江苏南京210019 [2]南京理工大学,江苏南京210094 [3]东南大学,江苏南京210096

出　　处：《现代交通技术》2017年第3期79-81,110,共4页Modern Transportation Technology

基　　金：国家自然科学基金青年资助项目(项目编号:51208261);教育部人文社会科学青年基金资助项目(项目编号:12YCZH062);中央高校基本科研业务费专项资金资助项目(项目编号:30920140132033)

摘　　要：根据交通标志牌的颜色特性,文章提出一种基于二维高斯HSV颜色模型的交通标志牌检测方法。采取限制对比度自适应直方图均衡化对采集到的图像进行预处理,增强图像的对比度;建立二维高斯HSV颜色模型对预处理后的图像进行分割,并进行形态学操作和边缘检测,从而实现交通标志牌的准确定位。A traffic sign detection method is presented based on two-dimensional Gaussian distribution HSV color model according to the color characteristics of traffic signs. Considering the impacts of illumination variation while collecting image information, the contrast limited adaptive histogram equalization method is applied to the image pre-processing to increase the image contrast. Then, a two-dimensional Gaussian distribution HSV color model is built by statistics and used to segment the pre-processed images. Finally after applying the morphology operations and border detection on the segmentation result, the goal of traffic sign location in natural background is achieved.

关 键 词：交通标志检测 自适应直方图 HSV颜色模型 形态学操作 

分 类 号：U491.52[交通运输工程—交通运输规划与管理]