Face Recognition in Subspaces University of Calgary卡尔加里大学子空间人脸识别.pptVIP

下载本文档

7
0
约2.83万字
约 59页
2017-02-14 发布于上海
举报
版权申诉

Face Recognition in Subspaces University of Calgary卡尔加里大学子空间人脸识别.ppt

1、原创力文档（book118）网站文档一经付费（服务费），不意味着购买了该文档的版权，仅供个人/单位学习、研究之用，不得用于商业用途，未经授权，严禁复制、发行、汇编、翻译或者网络传播等，侵权必究。。
2、本站所有内容均由合作方或网友上传，本站不对文档的完整性、权威性及其观点立场正确性做任何保证或承诺！文档内容仅供研究参考，付费前请自行鉴别。如您付费，意味着您自己接受本站规则且自行承担风险，本站不退款、不进行额外附加服务；查看《如何避免下载的几个坑》。如果您已付费下载过本站文档，您可以点击这里二次下载。
3、如文档侵犯商业秘密、侵犯著作权、侵犯人身权等，请点击“版权申诉”（推荐），也可以打举报电话：400-050-0827(电话支持时间：9:00-18:30)。
4、该文档为VIP文档，如果想要下载，成为VIP会员后，下载免费。
5、成为VIP后，下载本文档将扣除1次下载权益。下载后，不支持退款、换文档。如有疑问请联系我们。
6、成为VIP后，您将拥有八大权益，权益包括：VIP文档下载权益、阅读免打扰、文档格式转换、高级专利检索、专属身份标志、高级客服、多端互通、版权登记。
7、VIP文档为合作方或网友上传，每下载1次，网站将根据用户上传文档的质量评分、类型等，对文档贡献者给予高额补贴、流量扶持。如果你也想贡献VIP文档。上传文档

Face Recognition in Subspaces 601 Biometric Technologies Course Abstract Images of faces, represented as high-dimensional pixel arrays, belong to a manifold (distribution) of a low dimension. This lecture describes techniques that identify, parameterize, and analyze linear and non-linear subspaces, from the original Eigenfaces technique to the recently introduced Bayesian method for probabilistic similarity analysis. We will also discuss comparative experimental evaluation of some of these techniques as well as practical issues related to the application of subspace methods for varying pose, illumination, and expression. Outline Face space and its dimensionality Linear subspaces Nonlinear subspaces Empirical comparison of subspace methods Face space and its dimensionality Computer analysis of face images deals with a visual signal that is registered by a digital sensor as an array of pixel values. The pixels may encode color or only intensity. After proper normalization and resizing to a fixed m-by-n size, the pixel array can be represented as a point (i.e. vector) in a mn-dimensional image space by simply writing its pixel values in a fixed (typically raster) order. A critical issue in the analysis of such multidimensional data is the dimensionality, the number of coordinates necessary to specify a data point. Bellow we discuss the factors affecting this number in the case of face images. Image space versus face space Handling high-dimensional examples, especially in the context of similarity and matching based recognition, is computationally expensive. For parametric methods, the number of parameters one needs to estimate typically grows exponentially with the dimensionality. Often, this number is much higher than the number of images available for training, making the estimation task in the image space ill-posed. Similarly, for nonparametric methods, the sample complexity - the number of examples needed to represent the underlying distribution of data effic