基于视觉注意机制的实时红外小目标预检测  

Real-time Pre-detection for IR Small Target Based on Attention Mechanism

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6期7-9,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 (6 9875 0 0 5 )

摘　　要：在红外小目标检测中引入视觉注意机制 ,以计算量较小的预检测为代价减小了目标检测的区域 .在注意机制中采用非线性采样模型 ,并利用相邻采样点形成显著性图 .实验证明基于注意机制的预检测能有效地提高系统的计算效率 ,同时也具有一定的抗噪能力 。An attention mechanism appropriate for infrared(IR) small target detection is proposed, which is used as pre detection to limit detection in the smaller area. Nonlinear sampling model is adopted and saliency map is formed based on contrast of neighboring sampling nodes. Experiments are presented to demonstrate the improvement of speed of IR small target detection with pre detection. It is proved that the detection has reasonable ability of noise resistance and can be used in the small target detection in clutter background.

关 键 词：注意机制 小目标检测 实时图像处理 抗噪能力 生物视觉系统 预检测 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]