脉冲神经网络自适应侧连接的运动检测
Motion Detection Based on Adaptive Lateral Connection of Spiking Neuron Networks in Primary Vision Cortex
投稿时间:2014-01-15  修订日期:2014-01-15
DOI:
中文关键词: 脉冲神经网络  运动检测  侧连接  中心环绕作用
英文关键词: spiking neuron networks  motion detection  lateral connection  center surround interaction
基金项目:(61002046);国家自然科学(60972158)
作者单位E-mail
刘海华 中南民族大学 生物医学工程学院 lhh@mail.scuec.edu.cn 
庄如云 中南民族大学 生物医学工程学院  
摘要点击次数: 601
全文下载次数: 
中文摘要:
      运动检测研究是计算机视觉中的一个重要研究内容。由于对象运动的非刚性、对比度的差异性,以及环境的复杂性,给运动检测造成了很大困难。为此,根据视觉皮层脉冲神经网络中神经元相互作用的神经生物学机制,提出了模拟初级视皮层脉冲神经网络生物学属性的运动检测方法。该方法利用脉冲神经元之间的侧连接实现中心环绕的抑制和易化作用;并根据运动强度自适应地确定相互作用范围以及连接强度。实验结果表明,该方法不仅能较大程度地抑制噪声,而且能改善不同对比度对象的运动检测,从而提升了运动检测的性能。
英文摘要:
      Motion Detection is an important research in computer vision. Due to the non-rigid of motion objects, the differences of contrast and the complexity of the environment, it caused great difficulties to motion detection. Hence, according to the neurobiological mechanisms of neurons interaction of the spiking neural network in visual cortex, this paper presents a bio-inspired method to detect motion by simulating primary visual cortex. The method achieves center surround inhibitive and facilitative effect of the spiking neurons by using neurons lateral connection, and adaptively determines the range of the interaction and the strength of the connection according to the motion intensity. The experiment results show that the adaptive lateral connection mechanism can obviously suppress noise, and improves motion detection of different moving objects with different contrast, thus enhancing the performance of motion detection.
View Fulltext   查看/发表评论  下载PDF阅读器
关闭