基于HOG特征和TSVM算法的车标识别

Vehicle logo recognition based on HOG feature and TSVM algorithm

支持向量机(Support Vector Machine,SVM)是建立车标识别模型的主要智能方法之一。考虑SVM存在计算复杂度高和无法实现增量学习等问题,提出一种基于孪生支持向量机(Twin SVM,TSVM)增量学习算法,并结合HOG特征设计一种车标识别系统。首先利用特征检测结合仿射变换技术,实现车标的精准定位;然后提取车标图像HOG特征,并通过对矩阵的逆运算进行分解和重组,实现TSVM增量学习。最后利用车标数据集训练分类模型,实现对车标的分类。实验结果表明,文中提出的算法在车标数据集上实现了91.77%的识别率,优于其他几种识别算法,证明了文中提出算法的有效性。

Support vector machine(SVM)is one of the main intelligent methods to establish vehicle logo recognition model.Considering that SVM has problems such as high computational complexity and inability to realize incremental learning,an incremental learning algorithm based on Twin support vector machine(ILTSVM)is proposed,and an automatic vehicle logo recognition system is designed combined with HOG feature.Firstly,using feature detection combined with affine transformation technology,the accurate positioning of vehicle logo is realized.Then,the HOG feature is extracted from the vehicle logo image.By decomposing the inverse of the matrix and reorganizing the inverse of the matrix,the ILTSVM algorithm is formulated.Finally,the ILTSVM model is trained with the vehicle logo data set,which realizes the classification of vehicle logo.The experiment results show that the recognition rate of the ILTSVM on vehicle logo data set is 91.77%,which is higher than that of other recognition algorithms,proving the effectiveness of ILTSVM algorithm.