Computer Engineering and Applications ›› 2011, Vol. 47 ›› Issue (36): 203-207.

• 图形、图像、模式识别 • Previous Articles     Next Articles

Vehicle detection method conjoint with SVM and Gabor filter

AN Jiyao,OU Zhifang   

  1. College of Information Science and Engineering,Hunan University,Changsha 410082,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2011-12-21 Published:2011-12-21

结合SVM和Gabor参数优化的车辆检测

安吉尧,欧志芳   

  1. 湖南大学 信息科学与工程学院,长沙 410082

Abstract: A new vehicle detection method based on niche genetic algorithm conjoint with SVM and Gabor filter is presented.Fully taking account of the character that the model selection of SVM is close connection with the early feature extraction,the optimization of parameters of SVM and the Gabor filters are effectively combined.The fitness function is designed by using the classification accuracy,the penalty factor and the number of support vector.Training the model of SVM by using the niche genetic algorithm can effectively improve the classification accuracy of SVM,and reduce the number of features and the time complexity.Experimental results show that the average accuracy of vehicle detection is 98.675% and only 30 features are used.It is essential to meet the requirements of real-time.

Key words: niche genetic algorithm, Gabor feature, support vector machine, vehicle detection

摘要: 提出了一种基于小生境遗传算法的SVM和Gabor参数优化的车辆检测方法。充分考虑SVM的模型选择和前期的特征提取有着紧密联系的特点,将SVM的参数和Gabor的参数优化有效结合,利用分类精度、惩罚因子以及支持向量个数构造适应度函数,运用小生境遗传算法训练SVM模型,可有效提高SVM的分类精度,减少特征量的个数,降低时间复杂度。实验结果表明,车辆检测平均实验精度可达到98.675%,仅需30个特征量,基本可以满足实时性的要求。

关键词: 小生境遗传算法, Gabor特征, 支持向量机, 车辆检测