• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>Intelligent manufacturing systems 1998(IMS'98) >MONITORING THE QUALITY OF NON-WOVEN TEXTILES USING STOCHASTIC TEXTURE DISCRIMINATION
【24h】

MONITORING THE QUALITY OF NON-WOVEN TEXTILES USING STOCHASTIC TEXTURE DISCRIMINATION

机译：使用随机纹理判别监控非织造织物的质量

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

Several features have been proposed for stochastic texture image analysis. It has been a common belief that such features could also discriminate between stochastic textures. Our experiments show that several of the proposed features aer correlated and provide redundant texture representations. An effective feature set for texture discrimination was found, but such features are "expensive" to measure. It is shown that, taking advantage of the representation redundancy, these features can be estimated from other features readily mesurable, using a standard technique ("least mean square"). This leads to an improvement over the results obtained with a conventional pattern recognition technique for feature extraction and selection (i.e. "diagonalization").
机译：已经提出了一些用于随机纹理图像分析的特征。普遍认为，这些特征还可以区分随机纹理。我们的实验表明，一些建议的特征相互关联并提供了冗余的纹理表示。找到了用于纹理识别的有效特征集，但是这种特征的测量“昂贵”。结果表明，利用表示冗余，可以使用标准技术（“最小均方”）从易于测量的其他特征中估计这些特征。这导致对通过用于特征提取和选择（即“对角化”）的常规模式识别技术获得的结果的改进。

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