• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>Proceedings of the 4th International Conference on Intelligent Logistics Systems >New Algorithm for Face Recognition Using Small-area Edge Feature Fuzzy Fusion
【24h】

New Algorithm for Face Recognition Using Small-area Edge Feature Fuzzy Fusion

机译：小区域边缘特征模糊融合的人脸识别新算法

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Edge information is an important image feature. The topology of important organs (such as eyes) is mademuch clearer by the human facial edge image,meanwhile the noise is decreased to a great extent,and the levelof gray scale is relatively lower. So,the algorithm for face recognition based on edge feature has lowercomputation and higher recognition accuracy. The rain (or max) operator proposed by Nakagowa and Rosenfeldis used to extract and construct edge images of original facial images,then those edge images are divided intoseven small-area edge images. Owing to information independence of small-area facial edge images,softrecognition based on small-area edge images can be implemented firstly. The traditional SVD (singular valuedecomposition) operates directly on each small-area edge image respectively,then small-area SV (singularvalue) can be extracted and corresponding small-area SV eigenvector can be obtained. According to these small-area edge features,the membership grades of small-area test images versus small-area training images can bedetermined. The identity of an unknown facial image is determined by the fuzzy fusion which aggregates small-area edge features. The experiments show that the proposed method has the characteristics of higher recognitionrate and better robustness.
机译：边缘信息是重要的图像功能。人脸边缘图像使重要器官（如眼睛）的拓扑结构变得更加清晰，同时噪声大大降低，灰度等级相对较低。因此，基于边缘特征的人脸识别算法具有较低的运算量和较高的识别精度。由Nakagowa和Rosenfeldis提出的rain（或max）算子用于提取和构造原始人脸图像的边缘图像，然后将这些边缘图像分为七个小区域边缘图像。由于小区域人脸边缘图像的信息独立性，可以首先实现基于小区域人脸图像的软识别。传统的SVD（奇异值分解）分别直接作用于每个小区域边缘图像上，然后可以提取出小面积SV（奇异值）并获得相应的小区域SV特征向量。根据这些小区域边缘特征，可以确定小区域测试图像与小区域训练图像的隶属度。未知面部图像的身份由模糊融合确定，融合了小区域边缘特征。实验表明，该方法具有较高的识别率和较好的鲁棒性。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

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