激光与光电子学进展, 2016, 53 (10): 101503, 网络出版: 2016-10-12
融合局部二值模式和Hu矩特征的车型识别
Vehicle Type Recognition Based on Combining Local Binary Pattern and Hu Matrix Feature
机器视觉 车型识别 特征融合 融合局部二值模式 Hu矩 machine vision vehicle type recognition feature fusion local binary pattern Hu matrix
摘要
提出融合局部二值模式(LBP)和Hu矩特征的车型识别算法。LBP特征能够很好地对车辆局部纹理进行描述,Hu矩属于全局特征,反映了车辆的形状轮廓信息,将这两种互补特征结合能更好地表达车型特征。设计了融合特征的提取方法,并结合支持向量机分类器构建了车型识别系统。实验结果表明,融合算法比单一的特征算法性能更优,提高了车型识别率。
Abstract
The vehicle type recognition algorithm which combines local binary pattern (LBP) and Hu matrix feature is proposed. LBP feature can describe the local texture of vehicle. Hu matrix features are global features, and can reflect the shape contour information of the vehicle. The recognition algorithm of the fusion of LBP and Hu matrix can describe the vehicle type characteristics. The vehicle recognition system is constructed by using the proposed fusion feature extraction method and the support vector machine classifier. The fusion algorithm achieves better performance than the single feature algorithm, and improves the vehicles recognition rate.
李澄非, 陈新华. 融合局部二值模式和Hu矩特征的车型识别[J]. 激光与光电子学进展, 2016, 53(10): 101503. Li Chengfei, Chen Xinhua. Vehicle Type Recognition Based on Combining Local Binary Pattern and Hu Matrix Feature[J]. Laser & Optoelectronics Progress, 2016, 53(10): 101503.