首页> 外文期刊>International Journal of Applied Pattern Recognition >Editorial: Towards feature extraction in unconstrained image recognition environments
【24h】

Editorial: Towards feature extraction in unconstrained image recognition environments

机译:社论:在不受限制的图像识别环境中实现特征提取

获取原文
获取原文并翻译 | 示例
           

摘要

Recently, there has been an increased awareness and interest in using pattern recognition techniques for a wide range of problems in the areas of biometrics (Wang, 2012), agriculture (Bauckhage and Kersting, 2013), and document analysis (Bunke and Wang, 1997). While most existing methods provide very high recognition accuracies with images collected under laboratory conditions with specifically imposed constraints, they tend to perform poor in real-time unconstrained conditions (Kamila, 2015). In unconstrained recognition environments, the quality of features generated by most methods tends to different significantly between the training and the test sets. This is attributed to the unexpected increase in the natural variability (Piper, 1992; Chen et al., 2012) in the test sets often resulting from sensing errors, illumination changes, shape deformation of the objects, changes in object poses and several types of noise in the channel. Developing robust features to changes in different types of natural variability remains as one of the most important pattern recognition problem.
机译:最近,人们越来越意识到使用模式识别技术来解决生物识别(Wang,2012),农业(Bauckhage和Kersting,2013)以及文档分析(Bunke和Wang,1997)领域中的许多问题,并对此产生了兴趣。 )。尽管大多数现有方法在具有特殊施加约束的实验室条件下收集的图像都具有很高的识别精度,但它们在实时无约束条件下往往表现较差(Kamila,2015年)。在不受约束的识别环境中,大多数方法生成的特征的质量在训练和测试集之间趋于明显不同。这归因于测试集中自然可变性的意外增加(Piper,1992; Chen等,2012),通常是由于感测误差,光照变化,物体的形状变形,物体姿态的变化以及几种类型的物体引起的。通道中的噪音。为不同类型的自然变异性的变化开发鲁棒的特征仍然是最重要的模式识别问题之一。

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