[1]姚汉利,赵金金,鲍文霞.基于特征融合和字典学习的交通标志识别[J].计算机技术与发展,2018,28(01):51-55.[doi:10.3969/ j. issn.1673-629X.2018.01.011]
　Dictionary Learning.Traffic Sign Recognition Based on Feature Fusion and[J].Computer Technology and Development,2018,28(01):51-55.[doi:10.3969/ j. issn.1673-629X.2018.01.011]
点击复制

基于特征融合和字典学习的交通标志识别()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
28
期数:
2018年01期
页码:
51-55
栏目:
智能、算法、系统工程
出版日期:
2018-01-10

文章信息/Info

Title:
Traffic Sign Recognition Based on Feature Fusion and
文章编号:
1673-629X(2018)01-0051-05
作者:
姚汉利; 赵金金; 鲍文霞
安徽大学 电子信息工程学院,安徽 合肥 230601
Author(s):
Dictionary Learning
YAO Han-li,ZHAO Jin-jin,BAO Wen-xia
关键词:
交通标志识别; 融合稀疏; 广义典型相关分析; HOG; GIST; K-SVD
Keywords:
School of Electronics and Information Engineering; Anhui University; Hefei 230601; China
分类号:
TP39
DOI:
10.3969/ j. issn.1673-629X.2018.01.011
文献标志码:
A
摘要:
由于类别的多样性、内部结构的相似性以及外界环境等因素的影响,交通标志识别一直是人工智能与模式识别领域中的难题之一,而影响识别准确性的主要因素是特征的鉴别性与冗余性。 为了提高交通标志的识别准确性,提出了融合稀疏
表示的方法。 首先提取交通标志的 HOG 与 GIST 特征,再使用广义典型相关分析对提取的两个特征进行融合,融合得到的特征既保留了两个特征的有效信息,同时也增强了特征的鉴别性,但多特征的融合,难免会产生一定的冗余性。 在不降低特征鉴别性的前提下,为了减少其冗余性,最后使用 K-SVD 对其进行字典学习稀疏表示。 实验结果表明,交通标志的融合稀疏方法的效果明显优于大多数的识别方法,即使用线性 SVM 在 GTSRB 数据集上的分类准确率为 99.23％。
Abstract:
Because of the diversity of the category,the similarity of the internal structure and the influence of the external environment,traffic sign recognition is one of the most difficult problems in the field of artificial intelligence and pattern recognition. However,the main factors of influencing recognition accuracy are the discrimination and redundancy of feature. Therefore,in order to improve the recognition accuracy of traffic signs,we propose a fusion-sparse representation method. Firstly,the features of HOG and GIST are extracted;then the generalized
canonical correlation analysis is used to fuse the two features,and the fusion feature not only retains their effective information,but also enhances the discrimination. However,the fusion of multiple features will produce redundancy in a certain degree. In order to reduce the redundancy without lowering the discrimination,the K-SVD is used for dictionary learning sparse representation finally. The experiment shows that the method of fusion-sparse is superior to the most ones,which obtains the classification accuracy by 99. 23％ on the GTSRB dataset
with the linear SVM.

相似文献/References:

[1]汤智超,苏琳,何超,等. 导盲机器人的交通标志视觉识别技术研究[J].计算机技术与发展,2014,24(09):23.
　TANG Zhi-chao,SU Lin,HEChao,et al. Research on Traffic Sign Visual Recognition Technology of Guiding Robot[J].Computer Technology and Development,2014,24(01):23.
[2]王新美,丁爱玲,雷梦宁,等.基于 CNN 和 SVM 融合的交通标志识别[J].计算机技术与发展,2020,30(06):7.[doi:10. 3969 / j. issn. 1673-629X. 2020. 06. 002]
　WANG Xin-mei,DING Ai-ling,LEI Meng-ning,et al.Traffic Sign Recognition Based on Combination of CNN and SVM[J].Computer Technology and Development,2020,30(01):7.[doi:10. 3969 / j. issn. 1673-629X. 2020. 06. 002]
[3]王 杨,王 傲,许佳炜,等.基于双通道动态像素聚合的交通标志识别算法[J].计算机技术与发展,2023,33(06):41.[doi:10. 3969 / j. issn. 1673-629X. 2023. 06. 007]
　WANG Yang,WANG Ao,XU Jia-wei,et al.Traffic Sign Recognition Algorithm Based on Dual-channel Dynamic Pixel Polymerization[J].Computer Technology and Development,2023,33(01):41.[doi:10. 3969 / j. issn. 1673-629X. 2023. 06. 007]

更新日期/Last Update: 2018-03-12