基于人脸检测分析户外广告关注度  

Using face detection technology to analyze the attention rate of outdoor advertising

作　　者：袁宇[1,2] 张瑞[1,2] 王兴东[1,2] 

机构地区：[1]上海交通大学电子工程系图像通信与信息处理研究所,上海200240 [2]上海市数字媒体处理与传输重点实验室,上海200240

出　　处：《信息技术》2012年第3期17-19,23,共4页Information Technology

基　　金：国家自然科学基金项目(NSFC61071155)

摘　　要：户外广告需要统计其真实的关注度以更好地选择广告投放策略。文中提出了以观看广告的正面人脸数与持续时间来统计关注度。在OpenCV人脸检测函数库的基础上,提出了通过降分辨率、提取与分解有效像素、重叠检测、人脸生命周期模型等方法来改善检测的速度与精度。并给出了实验结果,分析了算法的优缺点,提供了改进的思路。People should statistics the real attention rate of outdoor advertising in order to have a better advertising strategy.This paper proposes a method to statistics the attention rate using the number and duration of frontal faces who are watching the advertisements.Based on the OpenCV face detection library,it brings up several ideas to improve the accuracy and speed,such as resolution reduction,valid pixels extraction and decomposition,overlap checking,face life-cycle model,etc.At last,it provides the experimental result,the advantages,disadvantages of the algorithm and gives out some ideas to improve the algorithm.

关 键 词：人脸检测 广告关注度 OPENCV 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]