Computer Engineering and Applications ›› 2008, Vol. 44 ›› Issue (14): 39-41.

• 理论研究 • Previous Articles     Next Articles

Linear SVM based construction of cascade detectors

AN Ping,WU Tao,HE Han-gen   

  1. School of Machatronics and Automation,National University of Defense Technology,Changsha 410073,China
  • Received:2007-08-30 Revised:2007-11-23 Online:2008-05-11 Published:2008-05-11
  • Contact: AN Ping

一种基于线性SVM的级联分类器的构造方法

安 平,吴 涛,贺汉根   

  1. 国防科学技术大学 机电工程与自动化学院 自动化研究所,长沙 410073
  • 通讯作者: 安 平

Abstract: To detect objects quickly,a new method is presented to construct a cascade of SVM classifiers.The classifier which contains several weak linear SVM classifiers is simple to understand and is extremely efficient.The learning problem of every node in the cascade structure is described as a new quadratic programming problem in the framework of SVM,which makes every linear classifier achieve very high detection rate but only moderate false positive rate.The real experiments show that this method enjoys good generalization capacity and much fast speed compared with the traditional SVMs.

Key words: cascade, object detection, Support Vector Machine(SVM)

摘要: 为了对目标进行快速的检测,提出了一种新的基于支持向量机的级联式分类器的构造方法。该级联分类器由若干个线性SVM弱分类器构成,结构简单,分类时间极快。针对级联结构中的每个节点的训练给出了一个新的SVM框架下的二次规划模型,这使得每个节点都有较高的正样本检测率和适当的负样本错检率。实际的实验结果表明,与经典非线性SVM分类器相比,这种分类器在保持SVM较强泛化性能的优点的同时,在检测效率方面更是具有明显的优势。

关键词: 级联, 目标检测, 支持向量机(SVM)