一种面向交通标志识别系统的图像识别算法

doi:10.16180/j.cnki.issn1007-7820.2019.07.012

摘要/Abstract

摘要：

目前各类面向交通标志识别系统的机器算法大多存在计算复杂度高、实时性差等问题,文中基于ELM模型,加入改进的PCA方法,提出了一种面向交通标志识别系统的PCA-ELM图像识别算法。该算法依次通过HOG特征的提取、改进的PCA方法降维、ELM模型的特征训练,实现交通标志图像的识别。经过实验测试,发现该算法能够较好地兼顾识别率和计算复杂度,符合交通标志识别系统图像识别的准确性与实时性要求,具有一定的实用价值。

关键词: 交通标志, 图像识别, HOG, PCA, ELM

Abstract:

Most of the machine algorithms for the traffic sign recognition system have problems of high computational complexity and poor real-time performance. Hence, based on the ELM model and the improved PCA method, an image recognition algorithm called PCA-ELM for the traffic sign recognition system was proposed. With the help of the extraction of HOG features, dimensionality reduction of improved PCA method, and the feature training of ELM model, the algorithm could realize the recognition of traffic sign images. Being tested and compared, it was found that the PCA-ELM algorithm had high recognition rate with suitable or even lower computational complexity, which well meets the accuracy and real-time requirements of image recognition for the traffic sign recognition system and had certain practical value.

Key words: traffic sign, image recognition, HOG, PCA, ELM

中图分类号:

TP393

张长青,杨楠. 一种面向交通标志识别系统的图像识别算法[J]. 电子科技, 2019, 32(7): 60-64.

ZHANG Changqing,YANG Nan. An Image Recognition Algorithm for Traffic Sign Recognition System[J]. Electronic Science and Technology, 2019, 32(7): 60-64.

图/表 5

图1

图2

图3

图4

表1

参考文献 16

[1]	张民 . 基于支持向量机的交通标志检测与识别[D]. 合肥:安徽大学, 2015.
	Zhang Min . Traffic sign detection and recognition based on support vector machine[D]. Hefei:Anhui University, 2015.
[2]	秦瑾, 张新峰 . 基于改进Shape Context的分层交通标志识别系统[J]. 计算机工程与设计, 2014,35(1):183-187.
	Qin Jin, Zhang Xinfeng . Layered traffic sign recognition system based on improved shape context[J]. Computer Engineering and Design, 2014,35(1):183-187.
[3]	王正玉, 李宁辉 . 基于SIFT算法的智能交通标志识别系统的设计与研究[J]. 微型机与应用, 2014,33(17):48-50.
	Wang Zhengyu, Li Ninghui . Design and research of intelligent traffic sign recognition system based on SIFT algorithm[J]. Microcomputer and Application, 2014,33(17):48-50.
[4]	Jeraj G . Traffic sign recognition system[J]. Journal of Physics Condensed Matter An Institute of Physics, 2011,23(34):345-403.
[5]	张志佳, 何纯静, 巩凯旋 , 等. 基于PC-LDA的交通标志形状特征识别方法[J]. 沈阳工业大学学报, 2014,36(5):550-554.
	Zhang Zhijia, He Chunjing, Gong Kaixuan , et al. Traffic sign shape feature recognition method based on PC-LDA[J]. Journal of Shenyang University of Technology, 2014,36(5):550-554.
[6]	麦海波, 苏时勇, 苏一峰 . 基于颜色分割和局部LPT变换的圆形标志检测算法[J]. 电子设计工程, 2016,24(21):191-193.
	Mai Haibo, Su Shiyong, Su Yifeng . Circular mark detection algorithms based on color segmentation and local LPT Transform[J]. Electronic Design Engineering, 2016,24(21):191-193.
[7]	邢晓敏, 刘威 . 雾天交通场景中单幅图像去雾[J]. 中国图象图形学报, 2018,21(11):1440-1447.
	Xing Xiaomin, Liu Wei . Single image defogging in foggy traffic scenes[J]. Chinese Journal of Image Graphics, 2018,21(11):1440-1447.
[8]	Li M, Yuan B . 2D-LDA:A statistical linear discriminant analysis for image matrix[J]. Pattern Recognition Letters, 2005,26(5):527-532.
[9]	Jin J, Fu K, Zhang C . Traffic sign recognition with hinge loss trained convolutional neural networks[J]. IEEE Transactions on Intelligent Transportation Systems, 2014,15(5):1991-2000. doi: 10.1109/TITS.2014.2308281
[10]	Roweis S . EM algorithms for PCA and SPCA[J]. Advances in Neural Information Processing Systems, 1997(10):626-632.
[11]	闫驰 . 基于PSO-BP神经网络的无线传感器网络定位算法[J]. 电子科技, 2016,29(4):56-58.
	Yan Chi . Location algorithm of wireless sensor networks based on PSO-BP neural network[J]. Electronic Science and Technology, 2016,29(4):56-58.
[12]	Stallkamp J, Schlipsing M, Salmen J , et al. Man VS. computer: Benchmarking machine learning algorithms for traffic sign recognition[J]. Neural Networks, 2012,32(2):323-332.
[13]	Stallkamp J, Schlipsing M, Salmen J, et al. The German traffic sign recognition benchmark: a multi-class classification competition [C].Boston:International Joint Conference on Neural Networks, 2011.
[14]	黄跃凯, 徐丹, 曾昊 , 等. 基于颜色对抗和旋转对称的路标检测算法[J].电子设计工程, 2014(19):169-172.
	Huang Yuekai, Xu Dan, Zeng Hao , et al. Road sign detection algorithms based on color countermeasure and rotational symmetry[J].Electronic Design Engineering, 2014(19):169-172.
[15]	陶陈, 徐丹, 张海玉 , 等. 基于视觉显著性和对称投影的禁止路标检测算法[J].计算机与现代化, 2013(2):35-37.
	Tao Chen, Xu Dan, Zhang Haiyu , et al. Prohibited landmark detection algorithms based on visual significance and symmetric projection[J].Computer and Modernization, 2013(2):35-37.
[16]	徐丹 . 自然场景下交通标志检测和分类算法研究[D]. 南京:南京理工大学, 2015.
	Xu Dan . Research on traffic sign detection and classification algorithm in natural scenes[D]. Nanjing: Nanjing University of Technology, 2015.

方法	Mandatory/%	Derestriction/%	Unique/%	Danger/%	Total/%
PCA-ELM	99.81	94.01	99.88	96.74	97.78
Committee of CNNs	99.91	99.63	99.13	98.99	99.39
Multi-scale CNNs	97.17	94.50	98.65	98.11	98.28
LDA on HOG 2	97.15	86.77	98.71	93.69	95.73
2LDA+5SVM	96.98	99.73	99.45	94.18	96.44