Performance enhancement by the improvement on the database for parallel optical facial recognition system

并行光学人脸识别系统数据库改进提高性能

基本信息

批准号：
10650045
负责人：
KODATE Kashiko
金额：
$ 2.24万
依托单位：
Japan Women's University
依托单位国家：
日本
项目类别：
Grant-in-Aid for Scientific Research (C)
财政年份：
1998
资助国家：
日本
起止时间：
1998 至 1999
项目状态：
已结题

来源：
https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-10650045/
关键词：
diffractive optical element parallel optical recognition system compact optical parallel correlator analog optical computer holographic multiple optical memory distributed database average face database application for security ホログラフィック多重光

项目摘要

The main object of this research is performance enhancement of parallel optical facial recognition by the construction of the new database on 3-dimentional wire flame model and the distributed database, and realization of multiple optical storage. The main results throughout these two years' research since 1998 are presented below.1) Recognition rate improvement by the construction of the database based on 3-dimentional wire flame model :We constructed two kinds of new database using the wire flame model. The one is to make the average face from one's several facial expressions and the other is to create some multiplex facial images, which contains five of each one. Choosing ten persons from the registered ones with some changes over time and the unregistered, we performed recognition experiments and concluded that these two database are conductive to recognition.2) Recognition speed improvement by construction of the distributed database :Categorization division by two kinds of facial … More index for the facial image retrieval was carried out from mass database. It was confirmed that the distribution to five groups was possible by two kinds of facial index. Reproducibility of 98% can be obtained for the images taken up 6 years ago and operation speed can be improved when the distributed database is used.3) Readout speed improvement by using the photo-refractive multiple optical memory :We sought to optimize these conditions with a view to applying SBN signal crystal to multi-channel multiplexed memory for reference image database of our facial recognition system and succeeded in recording 120 images.4) Application experiments to security system :The recognition of portraits, of which acquisition conditions very widely, is carried out. It was confirmed to have the robustness in which this system is excellent, on the assumption of the crime investigation. The result of the recognition experiments shows that the recognition rate exceeds 80% even after the portraits were taken a few years ago.5) Trial fabrication of compact parallel optical correlator :A compact prototype parallel optical correlator, whose size is 230×150×163 mmィイD13ィエD1, wad designed and devised. The experimental results of this compact system to facial recognition were evaluated with 100 person's database and proved to be successful. Less

这项研究的主要目的是通过在3维线火焰模型和分布式数据库上构建新数据库以及实现多个光学存储，通过构建新数据库来增强平行光学面部识别的性能。自1998年以来这两年的研究中的主要结果介绍了1）基于3维线火焰模型的数据库识别率提高了识别率：我们使用电线火焰模型构建了两种新数据库。一个是使一个人的几种面部表情中的平均面部表情，而另一个是创建一些多个面部图像，其中包含五个面部图像。我们从注册者那里选择十个人随着时间的流逝而进行了一些变化，并且没有注册，我们进行了识别实验，并得出结论，这两个数据库是识别的导电。2）通过构建分布式数据库的识别速度提高：分布式数据库分别通过两种面部面部的分类部门……面部图像检索的更多索引是从质量数据群中进行的。有人证实，两种面部指数可能可以分配给五组。 Reproducibility of 98% can be obtained for the images taken up 6 years ago and operation speed can be improved when the distributed database is used.3) Readout speed improvement by using the photo-refractive multiple optical memory :We sensed to optimize these conditions with a view to apply SBN signal crystal to multi-channel multiplexed memory for reference image database of our facial recognition system and succeeded in recording 120 images.4) Application experiments to security系统：对肖像的认识，对肖像的收购条件非常广泛。在犯罪调查的假设上，它证实具有鲁棒性，该系统是极好的。识别实验的结果表明，即使在肖像画中摄入肖像之后，识别率也超过80％。5）紧凑的平行光学相关器的试验制造：紧凑的原型平行光学相关器，其大小为230×150×163 mm，WAD设计和设计。通过100人的数据库评估了这种紧凑系统对面部识别的实验结果，并被证明是成功的。较少的