首页> 外国专利> OBJECT FEATURE EXTRACTION APPARATUS, OBJECT FEATURE EXTRACTION METHOD, OBJECT FEATURE EXTRACTION PROGRAM

OBJECT FEATURE EXTRACTION APPARATUS, OBJECT FEATURE EXTRACTION METHOD, OBJECT FEATURE EXTRACTION PROGRAM

机译:对象特征提取装置,对象特征提取方法,对象特征提取程序

摘要

PROBLEM TO BE SOLVED: To extract a feature of an object without separating an object from an image and without a need of a plurality of frame images and model information.;SOLUTION: An object feature extraction apparatus includes a feature extraction unit 210; a feature appearance frequency distribution calculation unit 220; and an object feature extraction unit 230. The feature extraction unit 210 extracts a plurality of feature sets from a plurality of image sets corresponding to a plurality of categories, respectively. The feature appearance frequency distribution calculation unit 220 calculates a plurality of feature appearance frequency distributions on the basis of the plurality of the feature sets, respectively. The object feature extraction unit 230 selects a plurality of important features from a plurality of features on the basis of comparison results of the plurality of the feature appearance frequency distributions, and calculates a plurality of object features indicating a plurality of frequencies where the plurality of the important features appear on the basis of the plurality of the feature appearance frequency distributions and the plurality of the important features.;COPYRIGHT: (C)2011,JPO&INPIT
机译:解决的问题:在不从图像分离对象并且不需要多个帧图像和模型信息的情况下提取对象的特征。解决方案:对象特征提取设备包括特征提取单元210;和特征提取单元。特征出现频率分布计算单元220;特征提取单元210分别从与多个类别相对应的多个图像集中提取多个特征集。特征出现频率分布计算单元220分别基于多个特征集来计算多个特征出现频率分布。对象特征提取单元230基于多个特征出现频率分布的比较结果,从多个特征中选择多个重要特征,并计算指示多个频率的多个对象特征,其中多个重要特征是根据多个特征出现频率分布和多个重要特征而出现的。; COPYRIGHT:(C)2011,JPO&INPIT

著录项

  • 公开/公告号JP2010282581A

    专利类型

  • 公开/公告日2010-12-16

    原文格式PDF

  • 申请/专利权人 NEC CORP;

    申请/专利号JP20090137626

  • 发明设计人 SENDA SHUJI;ISHITA TETSUO;

    申请日2009-06-08

  • 分类号G06T7/00;

  • 国家 JP

  • 入库时间 2022-08-21 18:21:17

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