基于支持向量机的彩色人脸检测技术

【作者】 曹德平；

【导师】 陈锻生；

【作者基本信息】 华侨大学 ， 计算机应用技术， 2005， 硕士

【摘要】 人脸检测是计算机视觉、模式识别研究中一个很受关注的研究热点,而且它在自动人脸识别系统、视觉监控、基于内容的检索以及新一代的人机交互技术等领域都有着广泛的应用。然而,由于人脸检测问题本身的复杂性,使得人脸检测算法一般都存在计算量大、速度慢、误报率高的弱点。 支持向量机(Support Vector Machines,简称SVM)是建立在统计学习理论基础上的一种新的模式识别技术。通过采用结构风险最小化原则,SVM较好的解决了小样本学习的问题。利用核函数思想,它又能把到高维特征空间中的映射和特征空间中的点积简化为输入空间中的解析式计算,降低了算法的复杂度。然而,SVM仍然存在着一些问题,如训练算法速度慢、算法复杂以及检测阶段运算量大等。 本文针对彩色图像提出了一种新的快速正面人脸检测方法。首先采用肤色检测技术获得图像中可能存在人脸的区域,然后采用基于SVM的检测方法在可能存在人脸的区域进行检测。因为检测的区域比较小,所以该方法在保证检测的准确性同时获得了接近实时的检测速度。基于SVM的方法包括训练和检测两部分:训练过程是用大量人脸样本、“非人脸”样本训练SVM分类器,使之获得一个最优分类超平面:检测阶段用训练好的SVM分类器检测图像中的人脸。 试验证明本文提出的人脸检测方法是合理的,具有一定的实用价值。更多还原

【Abstract】 Face detection has been a research focus of computer vision and pattern recognition for a long time. It also has the application in a wide range of fields such as automatic face recognition systems, visual surveillance, content-based retrieval, advanced human and computer interaction. But, due to the complexity of the purpose, most face detection methods have the weakness of large computation, low efficiency and many false reports among the detection result.Support Vector Machines (SVM) is a new pattern recognition technology that is established on Statistical Learning Theory. It can solve small-sample learning problems better by using Experiential Risk Minimization. Moreover, this theory can change the problem in non-linearity space to that in the linearity space in order to reduce the algorithm complexity by using the kernel function idea. However, SVM theory performance has been validated in many practical applications, there are still some drawbacks. For example: train speed is slow, algorithm is complex and check phase operation is large, etc.This paper proposed a new method to detect frontal view human faces in color images quickly. This dissertation adopting background separate technique with skin color detection technique combine together to get the area which may be have person’s face, then adoption SVM method to proceed detection. Because of the area is small, this kind of method namely guaranteed the detection rate at the same time detections peed that near to the real time. SVM-Based Face Detection made of training section and detecting section. A lot of face samples and "not face" samples are used to train the SVM classifier, to get optimal separating hyper plane in the training. And SVM classifier is used to detect faces in the detecting.Examination in some image databases prove this method have veryhigh detection rate, and have the very high theories value and practical value.更多还原