多尺度视黄线在人脸识别分析中的应用

IF 0.5 Q4 COMPUTER SCIENCE, THEORY & METHODS

JOURNAL OF INTERCONNECTION NETWORKS Pub Date : 2020-12-08 DOI:10.15575/JOIN.V5I2.668

S. Supriyanto, M. Harika, Maya Sri Ramadiani, D. R. Ramdania

{"title":"多尺度视黄线在人脸识别分析中的应用","authors":"S. Supriyanto, M. Harika, Maya Sri Ramadiani, D. R. Ramdania","doi":"10.15575/JOIN.V5I2.668","DOIUrl":null,"url":null,"abstract":"The main challenge that facial recognition introduces is the difficulty of uneven lighting or dark tendencies. The image is poorly lit, which makes it difficult for the system to perform facial recognition. This study aims to normalize the lighting in the image using the Multiscale Retinex method. This method is applied to a face recognition system based on Principal Component Analysis to determine whether this method effectively improves images with uneven lighting. The results showed that the Multiscale Retinex approach to face recognition's correctness was better, from 40% to 76%. Multiscale Retinex has the advantage of dark facial image types because it produces a brighter image output.","PeriodicalId":53990,"journal":{"name":"JOURNAL OF INTERCONNECTION NETWORKS","volume":"17 1","pages":"217-226"},"PeriodicalIF":0.5000,"publicationDate":"2020-12-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Multiscale Retinex Application to Analyze Face Recognition\",\"authors\":\"S. Supriyanto, M. Harika, Maya Sri Ramadiani, D. R. Ramdania\",\"doi\":\"10.15575/JOIN.V5I2.668\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"The main challenge that facial recognition introduces is the difficulty of uneven lighting or dark tendencies. The image is poorly lit, which makes it difficult for the system to perform facial recognition. This study aims to normalize the lighting in the image using the Multiscale Retinex method. This method is applied to a face recognition system based on Principal Component Analysis to determine whether this method effectively improves images with uneven lighting. The results showed that the Multiscale Retinex approach to face recognition's correctness was better, from 40% to 76%. Multiscale Retinex has the advantage of dark facial image types because it produces a brighter image output.\",\"PeriodicalId\":53990,\"journal\":{\"name\":\"JOURNAL OF INTERCONNECTION NETWORKS\",\"volume\":\"17 1\",\"pages\":\"217-226\"},\"PeriodicalIF\":0.5000,\"publicationDate\":\"2020-12-08\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"JOURNAL OF INTERCONNECTION NETWORKS\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.15575/JOIN.V5I2.668\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q4\",\"JCRName\":\"COMPUTER SCIENCE, THEORY & METHODS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"JOURNAL OF INTERCONNECTION NETWORKS","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.15575/JOIN.V5I2.668","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"COMPUTER SCIENCE, THEORY & METHODS","Score":null,"Total":0}

引用次数: 0

摘要

面部识别带来的主要挑战是光线不均匀或黑暗趋势的困难。图像光线很差，这使得系统难以进行面部识别。本研究旨在利用多尺度Retinex方法对图像中的光照进行归一化。将该方法应用到基于主成分分析的人脸识别系统中，判断该方法是否能有效改善光照不均匀的图像。结果表明，采用多尺度Retinex方法进行人脸识别的正确率从40%提高到76%。多尺度Retinex具有深色面部图像类型的优势，因为它产生更亮的图像输出。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

查看原文本刊更多论文

Multiscale Retinex Application to Analyze Face Recognition

The main challenge that facial recognition introduces is the difficulty of uneven lighting or dark tendencies. The image is poorly lit, which makes it difficult for the system to perform facial recognition. This study aims to normalize the lighting in the image using the Multiscale Retinex method. This method is applied to a face recognition system based on Principal Component Analysis to determine whether this method effectively improves images with uneven lighting. The results showed that the Multiscale Retinex approach to face recognition's correctness was better, from 40% to 76%. Multiscale Retinex has the advantage of dark facial image types because it produces a brighter image output.

求助全文

通过发布文献求助，成功后即可免费获取论文全文。去求助

来源期刊

JOURNAL OF INTERCONNECTION NETWORKS COMPUTER SCIENCE, THEORY & METHODS-

自引率

14.30%

发文量

121

期刊介绍： The Journal of Interconnection Networks (JOIN) is an international scientific journal dedicated to advancing the state-of-the-art of interconnection networks. The journal addresses all aspects of interconnection networks including their theory, analysis, design, implementation and application, and corresponding issues of communication, computing and function arising from (or applied to) a variety of multifaceted networks. Interconnection problems occur at different levels in the hardware and software design of communicating entities in integrated circuits, multiprocessors, multicomputers, and communication networks as diverse as telephone systems, cable network systems, computer networks, mobile communication networks, satellite network systems, the Internet and biological systems.