摘要
提出一种人脸检测与识别的方法。运用Adaboost算法对输入的图像进行人脸检测并完成尺寸和光照的归一化,对训练样本集进行处理,运用主成分分析和Fisher脸判别实现对训练样本集空间的降维和分类,对归一化后的输入图像进行PCA转换和FLD投影,将得到的向量和训练样本集进行比较,从而分类。实验表明,本方法可以达到理想的识别效果。
Proposes a method of face detection and recognition. Inputs the images anduses Adaboost algorithm for face detection, completes normalization of size and illumination, dea and Fisher tls with the training samples, uses principal component analysiso realize the dimensionality reduction and classification of the training samples, does PCA conversion and FLD projection for normalized image, compares the vector and training samples in order to classify. Experiment shows, this method can achieve the desired recognition effect.
出处
《现代计算机》
2007年第5期50-52,共3页
Modern Computer
