基于组合矩特征和PTSVM的交通标志识别  

Traffic Sign Recognition Based on Combination Moment Features and SVM

作　　者：邓琼 何星 DENG Qiong;HE King(Anhui Polytechnic University,Wuhu 241000,China;Yangzhou University,Yangzhou 225000,China)

机构地区：[1]安徽工程大学电气工程学院,安徽芜湖241000 [2]扬州大学农学院,江苏扬州225000

出　　处：《洛阳理工学院学报（自然科学版）》2018年第2期51-56,共6页Journal of Luoyang Institute of Science and Technology：Natural Science Edition

摘　　要：针对交通标志识别算法识别精度低、计算复杂度高等问题,提出一种采用组合矩特征分类交通标志的方法。首先对图像进行预处理,并基于颜色特征分割,再采用形态学的方法处理粗分割图像,从而提取目标区域。然后分别提取标志图像的Legendre矩及小波矩特征,利用串行特征融合技术获取组合优化特征,将特征值输入渐进直推式支持向量机,并采用模拟退火算法对其进行参数优化,最后使用优化后的识别算法实现目标图像的识别。仿真结果表明,与现有的其他交通标志识别算法相比,采用组合矩与优化后渐进直推式支持向量机的识别方法有更好的识别效果。Aiming at the problems such as low recognition accuracy and high computation cost,which exist in the traditional traffic sign recognition algorithm,a method using combination moment feature is proposed. First of all,after preprocessing the image,with color feature segmentation and morphological method,the object region is obtained by processing the rough segmentation image. Extracting the Legendre moments and wavelet invariant moments of the image separately,the combinatorial optimization features are obtained by using the serial feature fusion technique. Then,the feature values are input to the progressive direct support vector machine and the simulated annealing algorithm is applied to the progressive direct support vector machine parameters optimization. Finally,the optimized algorithm is used to achieve the target image recognition. The simulation results show that the presented method can get a better recognition effect.

关 键 词：交通标志识别 LEGENDRE矩 小波不变矩 支持向量机 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TN919.8[自动化与计算机技术—计算机科学与技术]