摘要
本文通过比较交通标志检测和分类算法,利用阈值分割和神经网络算法思想构建了一个交通标志识别模型,该模型对于GTSRB数据集上的交通标志识别图片识别错误率能控制在5%以内,处理每帧的识别过程在150ms左右,较好地实现了交通标志的实时检测与分类。为相关交通管理部门提供了一套方便的管理技术。
In this paper,by comparing the traffic sign detection and classification algorithm,using the ideas of threshold segmentation algorithm and neural network,researchers constructed a traffic sign recognition model,the model for GTSRB data sets on traffic sign recognition image recognition error rate can be controlled within5%,the processing of each frame recognition process is controlled in150ms,implement the real-time detection and classification of traffic signs well.The researchers provided a convenient management technology for related traffic management department.
作者
徐彬森
魏元周
毛光明
李曼曼
XU Bin-sen;WEI Yuan-zhou;Mao Guang-ming;LI Man-man(BeiHang University (Beijing) Software College, Beijing 100191, China;Henan University of Finance and Economics (Zhengzhou) Computer and Information Engineering College, Zhengzhou 450046, China)
出处
《软件》
2017年第11期74-81,共8页
Software
关键词
交通标志识别
阈值分割
神经网络
Traffic sign detection
Threshold segmentation
Convolutional neural network (CNN)