实时红外小目标预检测中注意机制分析  

Analysis for Attention Mechanism in Real-time IR Small Target Pre-detection

作　　者：王岳环[1] 张天序[1] 

机构地区：[1]华中科技大学图像识别与人工智能研究所图像信息处理与智能控制教育部重点实验室

出　　处：《华中科技大学学报（自然科学版）》2001年第11期53-55,共3页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目 ( 69875 0 0 5 ) .

摘　　要：针对用于红外小目标预检测的基于局部对比度的注意机制 ,分析了其非线性采样模型参数与预检测生成的显著性图中目标 /背景信噪比的关系 ,为模型的参数选择提供了理论依据 .同时 ,在分析预检测算法计算量的基础上 ,比较了采用该预检测算法和不采用预检测算法的小目标检测之间的计算量差异 。According to the local contrast based attention mechanism in infrared (IR) small target pre detection, the analysis is proposed for the relationship between the key parameters of the nonlinear sampling model and the SNR(signal to noise ratio)in the saliency map generated, which is the basis of the parameters selection. The comparison between the computational complexity of small target detection with and without pre detection is also proposed based on the analysis of the computational complexity of the pre detection itself, which can demonstrate how the pre detection can enhance the efficiency of small target detection.

关 键 词：图像处理 计算效率 信噪比 红外小目标预检测 注意机制 非线性采样模型 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TN215[自动化与计算机技术—计算机科学与技术]