首页> 外国专利> Methods and systems for performing facial recognition, to train a model with reference to calculate a value of the threshold of similarity to a reference face model and to optimize an image for use in facial recognition

Methods and systems for performing facial recognition, to train a model with reference to calculate a value of the threshold of similarity to a reference face model and to optimize an image for use in facial recognition

机译:用于执行面部识别,用于参考训练模型以计算与参考面部模型的相似性阈值的值以及优化用于面部识别的图像的方法和系统

摘要

Methods and systems for performing facial recognition, to train a model with reference to calculate a value of the threshold of similarity to a reference face model and to optimize an image for use in facial recognition.The invention describes a method to perform recognition of face, which method comprises the steps of generating a model of the face middle (M ~ AV ~) - comprising a matrix of states representing regions of the face from a number of different face images (I ~ 1 ~ 1 ~ 2 ~,..., I. ~ J ~) and tre Inar model face of reference (M ~ 1 ~ m ~ 2 ~,..., m ~ n ~) for each of a number of faces known, where the model of face of reference (M1, M2, m3) is based on the model of face middle (MAV). A test image (it) is acquired for a face to be identified, and a best path through the model of face middle (M ~ AV ~) is calculated, based on the test image (I ~ t ~).A degree of similarity is evaluated for each type of face of reference (M ~ 1 ~ m ~ 2 ~,..., m ~ n ~) in relation to the test image (I ~ t ~) by applying the best path of model face middle (M ~ AV ~) to each type of face of reference (M ~ 1 ~ m ~ 2 ~,..., m ~ n ~) to identify the model of the face of re Reference (M ~ 1 ~ m ~ 2 ~,..., m ~ n ~) is most similar to the test image (I ~ t ~), the model of face of reference identified (M ~ 1 ~ m ~ 2 ~,..., m ~ n ~) is subsequ00faentemente accepted or rejected based on their degree of similarity. In addition, the invention describes a system to perform recognition of face.Also, the invention describes a method and system for training a model of face of reference (M ~ 1 ~) that can be used in the face recognition system, a method and system for calculating a value of the threshold of similarity to a reference face model (M ~ n ~) that may be used in a Face recognition system, and a method and a system for optimizing images (I, R ~ t ~ I ~ t ~ g ~ 1 ~,G ~ 2 ~,..., G ~ J ~ t ~ 1 ~ t ~ 2 ~ t ~ m ~ t ~ NEW ~) that can be used in face recognition system.
机译:用于执行面部识别的方法和系统,用于参考训练模型以计算与参考面部模型的相似性阈值并优化用于面部识别的图像。本发明描述了一种用于执行面部识别的方法,哪种方法包括生成脸部中间模型(M〜AV〜)的步骤-包括代表多个不同脸部图像(I〜1〜1〜2〜,...的脸部区域的状态矩阵) ,I.〜J〜)和tre Inar模型参考面(M〜1〜m〜2〜,...,m〜n〜)为已知的多个面中的每一个,其中参考面的模型( M1,M2,m3)基于面部中间(MAV)模型。获取要识别的面部的测试图像(it),并根据测试图像(I〜t〜)计算出通过面部中间模型(M〜AV〜)的最佳路径。通过应用模型人脸中间的最佳路径(M〜1〜m〜2〜,...,m〜n〜)对与测试图像(I〜t〜)相关的每种参考脸进行评估M〜AV〜)以每种类型的参考脸(M〜1〜m〜2〜,...,m〜n〜)来识别重新参考的脸部模型(M〜1〜m〜2〜 ,...,m〜n〜)与测试图像(I〜t〜)最相似,识别出参考人脸模型(M〜1〜m〜2〜,...,m〜n〜)根据其相似程度,subsubeq u00faentemente被接受或拒绝。另外,本发明描述了用于执行面部识别的系统。此外,本发明还描述了用于训练可以在面部识别系统中使用的参考面部(M〜1〜)的模型的方法和系统。可以在面部识别系统中使用的与参考面部模型(M〜n〜)的相似度阈值的计算系统,以及用于优化图像(I,R〜t〜I〜t的方法和系统) 〜g〜1〜,G〜2〜,...,G〜J〜t〜1〜t〜2〜t〜m〜t〜NEW〜可用于人脸识别系统。

著录项

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号