作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2011, Vol. 37 ›› Issue (22): 151-152. doi: 10.3969/j.issn.1000-3428.2011.22.049

• 人工智能及识别技术 • 上一篇    下一篇

视觉感知中特征捆绑建模方法的研究

李海芳,尹 清   

  1. (太原理工大学计算机科学与技术学院,太原 030024)
  • 收稿日期:2011-06-20 出版日期:2011-11-18 发布日期:2011-11-20
  • 作者简介:李海芳(1963-),女,教授、博士,主研方向:智能信息处理,情感计算;尹 清,硕士研究生
  • 基金资助:

    国家自然科学基金资助项目(61070077);山西省自然科 学基金资助项目(2010011020-2)

Research on Modeling Method of Feature Binding in Visual Perception

LI Hai-fang, YIN Qing   

  1. (College of Computer Science and Technology, Taiyuan University of Technology, Taiyuan 030024, China)
  • Received:2011-06-20 Online:2011-11-18 Published:2011-11-20

摘要: 针对视觉感知中特征的捆绑问题,在传统脉冲耦合神经网络(PCNN)的基础上,提出一种基于强度的PCNN模型。在该模型中,神经元邻域内脉冲发放总强度将不同的特征分离开来,而神经元自身的脉冲发放强度又将属于同一感知对象的不同特征捆绑起来。仿真结果证明,该模型可以实现特征的分离和捆绑。

关键词: 特征分离, 特征捆绑, 视觉感知, 脉冲耦合神经网络

Abstract: In view of feature binding in visual perception, this paper presents an advanced Pulse Coupled Neural Network(PCNN) model based on traditional PCNN. In this model, the pulse of the total strength in the neighborhood of neuron makes different characteristics separate, while the pulse of strength of neuron itself makes different characteristics bind. Simulation result indicates that the model can achieve feature separation and binding, so that it accomplishes the task of feature binding in visual perception and provides certain reference value for further research of feature binding in visual perception.

Key words: feature separation, feature binding, visual perception, Pulse Coupled Neural Network(PCNN)

中图分类号: