基于LARK特征提取的AdaBoost人脸检测系统

AdaBoost Face Detection Based on LARK Feature Extraction

由于图像易受到外部条件及图像背景的影响,AdaBoost人脸检测方法在特征分类的过程中,单个分类器存在将人脸图像误判为非人脸图像的情况,致使分类器在检测过程中存在误差。文中通过研究AdaBoost人脸检测方法,并将LARK特征提取方法应用到特征分类当中,使原有的方法上得到改进,从而有效提高了人脸检测的准确度。

As the images are vulnerable to external conditions and image background in the imaging, some nonface images are classified as face images by a single classifier in AdaBoost face detection. And for feature classification, it is difficuh to detect face images by using a single classifier. This paper studies the AdaBoost face detection and uses LARK for feature extraction to construct a face system, which efficiently improves the accuracy of face detection.