基于注意覆盖的感兴趣区域增强策略  

Enhancement Strategy for Region of Interest Based on Attentional Shroud

作　　者：孟祥林[1] 王正志[1] 

机构地区：[1]国防科技大学机电工程与自动化学院

出　　处：《计算机工程》2011年第4期18-20,共3页Computer Engineering

基　　金：国家自然科学基金资助项目(60835005);国防科技大学科研计划基金资助项目(JC09-03-04)

摘　　要：提出一种基于注意覆盖的感兴趣区域增强策略。将感知表面作为注意选择的基本单元,将自顶向下的注意信号引入表面填充机制,利用非线性扩散机制在感知表面形成一种形状拟合分布,使得被注意的感知物体活性得到增强。仿真结果表明,该策略可以有效地增强感兴趣区域,具有神经生理和心理学合理性,输出结果可用于区域分割、目标识别和场景分析。A strategy for enhancing region of interest with attentional shroud is proposed. Perceptual surfaces are assumed to be units of selective attention. Top-down attentional signals are introduced into surface filling-in mechanisms. Perceptual surface generates a form-fitting distribution of spatial attention through nonlinear diffusion. Consequently the activity of attentive perceptual object surface is enhanced. Simulation results show that the proposed strategy can enhance the interested region efficiently. It is physiologically and psychologically plausible and its results can be applied to image segmentation, object recognition and scene analysis.

关 键 词：注意覆盖 感兴趣区域 感知表面 填充 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]