Android平台实时人脸检测和性别识别的研究与实现  

Research and Implementation of Real-time Face Detection and Gender Recognition on Android Platform

作　　者：余彧[1] 陆亨立[1] 陆小峰[1] 李莹娇 

机构地区：[1]上海大学通信与信息工程学院,上海200072

出　　处：《电视技术》2014年第13期186-190,共5页Video Engineering

摘　　要：提出一种基于开源计算机视觉库OpenCV(Open Source Computer Vision Library)实现Android平台中实时人脸检测和性别识别的方法,阐述了在运行Android系统的嵌入式平台中采用AdaBoost算法实现人脸检测以及FisherFace算法实现性别识别的原因和具体实现方法,并分别在基于ARM和X86构架的两个嵌入式平台中进行了大量实验和测试数据对比。实验结果表明Android平台中能够实时地进行人脸检测和性别识别,且获得96.35%的人脸检测准确率和82.64%的性别识别准确率。A real-time face detection and gender recognition method on Android platform using OpenCV is proposed. Why and how to using AdaBoost algorithm for face detection and FisherFace algorithm for gender recognition on embedded platforms which are running the Android operating system is elaborated. A large number of experiments and test data comparisons on two embedded platform based on ARM and X86 architecture respectively are con- ducted. The experimental results prove that real-time face detection and gender recognition can be realized on Android platform. The face detection accuracy is 96.35%, arid the gender recognition accuracy is 82.64%.

关 键 词：Android 人脸检测 性别识别 ADABOOST FISHERFACE 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]