基于AdaBoost和LLE的视频人脸特征提取研究

AdaBoost and LLE for face feature extraction techniques in video sequence

特征提取是生物特征识别的关键环节.本文提出了一种基于AdaBoost和LLE的视频人脸特征提取方法.其大致思路是用VFW技术对视频图像进行采集、接着将AdaBoost算法对采集的图像进行人脸检测,最后应用LLE算法对检测到的人脸图像进行降维并提取出特征.项目实践表明,该方法具有便捷性、实用性和有效性.

Feature extraction is a key component of biometrics. This paper puts forward a scheme based on AdaBoost and LLE from the video sequences. The basic idea for the schemes is that frame images is captured from video sequence by FVW, face detection for that images is carried out according to AdaBoost algorithm, and Dimensionality Reduction and feature extraction for Located face image is implemented by LLE. Experimental results indicate that this scheme presented in this paper is convenient, practicable and available.