融合PCA的支持向量机人脸检测研究

支持向量机（Support Vector Machine,SVM）作为一种经典的非线性分类器，用于模式识别，可以将训练样本从不可线性分类的低维空间映射到可线性分类的高维空间，再做分类，本文主要训练支持向量机使它学会区分人脸和非人脸。支持向量机的数学推导完备，算法逻辑严密，整体上比Adaboost算法复杂，但在样本量较少的情况下效果良好，因此有样本优势。支撑它的理论包含泛化性理论、最优化理论和核函数等，这些理论也被学术界广泛用于其他机器学习算法如神经网络，几十年来被证明具有很高的可靠性。同时本文论述主成分分析技术（PCA）用于压缩数据，实现数据降维，在数据预处理方面算法提供了很大帮助，使SVM支持向量机的输入数据维数大幅下降，大大提高了运算和检测时间。

Research on face detection based on improved support vector machine combined with PCA

Support Vector Machine (SVM), as a classical nonlinear classifier, is used for pattern recognition. It can map the training samples from the low dimensional space of never linear classification to the high dimensional space of the linear classification, and then do the classification. This paper mainly trains the classifier to make it learn to distinguish the face and the non face. The support vector machine has complete mathematical derivation, rigorous algorithm logic, and more complex than Adaboost on the whole, but it has a good effect in the case of less sample size, so there is a sample advantage. The theory that supports it includes generalization theory, optimization theory and kernel function, which are also widely used in other machine learning algorithms, such as neural networks, which have been proved to have high reliability for several decades. At the same time, this paper discusses the principal component analysis (PCA) technology to compress data, realize data reduction, and provide great help to the algorithm of data preprocessing, which greatly reduces the dimension of input data of SVM support vector machine, and greatly improves the operation and detection time.