• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>International Journal of Innovative Computing Information and Control >LEARNING REGULARIZED MULTI-VIEW STRUCTURED SPARSE REPRESENTATION FOR IMAGE ANNOTATION
【24h】

LEARNING REGULARIZED MULTI-VIEW STRUCTURED SPARSE REPRESENTATION FOR IMAGE ANNOTATION

机译：用于图像标注的学习调节多视图结构化稀疏表示

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

Automatic image annotation is an mportant problem in computer visionouing to its critical role in image retrieval. In order to exploit the diversities of different features n a sample as well as the similarities, we present a regularized multi-uieustructured sparse representation model for image annotation. In this model, handcraftedvisual features, deep learning based features and label information are considered as dferent views. Each view is coded on its associated dictionary to allow flexibility of codingcoefficients from different views, while the disagreement between each view and a soft-consensus regularization term is minimized to keep the similarity among multiple viewsThe weight for each vieu is learned in the coding stage, and a weighted label predictionand propagation method is also proposed. Eaperimental results on ESP Game and IAPRTC-12 datasets demonstrate the effectiveness of the proposed approach compared withother related approaches for image-annotation task.
机译：自动图像标注是计算机视觉中的重要问题，这是由于其在图像检索中的关键作用。为了利用样本中不同特征的多样性以及相似性，我们提出了一种用于图像标注的正则化多结构稀疏表示模型。在此模型中，手工制作的视觉功能，基于深度学习的功能和标签信息被视为不同的视图。每个视图都在其关联的字典上进行编码，以允许来自不同视图的编码系数的灵活性，同时最小化每个视图与软共识正则化术语之间的分歧，以保持多个视图之间的相似性。提出了加权标签预测与传播方法。 ESP游戏和IAPRTC-12数据集的实验结果证明，与其他相关方法相比，该方法的有效性。

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