摘要
为配合刑事侦查工作和公安网上追逃的迫切需要,对人脸模拟画像的检索作初步研究。首先,根据网上追逃数据库中的人脸照片设计出一种半自动的归一化算法。其次,研究了人脸模拟画像和人脸照片的边缘检测方法,给出一种自适应的人脸照片边缘检测算法,从而得到三种用于人脸模拟画像检索的统计特征。最后,建立了人脸模拟画像检索的实验数据库,给出相似性量度,并对实验数据进行了分析。为构建实用的人脸模拟画像检索系统的开发提供了理论支持;对金盾工程的完善起到良好的促进作用。
This paper aims to cope with the criminal investigation work and to meet the exigency of cyber pursuit by the Police, and the retrieval of simulated face portrait is preliminarily studied. Firstly we designed a semi-automatic normalization algorithm based on the face ima- ges in database of China Cyber Pursuit. Secondly,the simulated face portrait and the edge detection method of face photo were studied and a self-adaptive edge detection algorithm of the face photo was presented, from which three statistical features for retrieval of the simulated face portrait were exacted. At last we established an experimental retrieval database of the simulated face portrait and gave out the measurement of similarity, as well as analysed the experimental data. The study would provide a theoretical support to the development of constructing a practical retrieval system of the simulated face portrait, and is expected to well promote the amelioration of the Golden Shield Project.
出处
《计算机应用与软件》
CSCD
2009年第8期238-240,246,共4页
Computer Applications and Software
关键词
人脸模拟画像
人脸照片
人脸识别
边缘检测
统计特征
Simulated face portrait Face photo Face recognition Edge detection Statistical features
