红外序列图像中运动小目标的检测方法  

Detection Method of Small Moving Targets in IR Sequence

作　　者：李国宽[1] 彭嘉雄[1] 李红[2] 

机构地区：[1]华中理工大学模式识别与人工智能研究所图像信息处理与智能控制国家教育部重点实验室 [2]华中理工大学数学系

出　　处：《华中理工大学学报》1999年第11期25-27,共3页Journal of Huazhong University of Science and Technology

基　　金：国家自然科学基金

摘　　要：研究红外图像中运动小目标的检测和跟踪问题．根据红外图像中背景干扰和噪声的特点，采用梯度倒数加权低通滤波器对图像进行滤波．在目标分割和标记后，沿目标轨迹进行能量积累以搜索和跟踪小目标．实验结果表明，该方法能极大地提高实际红外目标图像的信噪比，改善目标分割效果，能有效地进行小目标的检测和跟踪，尤其适用于目标遮掩或偶尔出现目标丢失的情况．A new algorithm is presented for detecting and tracking small moving targets in IR image sequences. Corresponding with characteristics of background clutter and noise, the gradient inverse weighted neighborhood averaging (GIWNA) filter is adopted. The energy accumulation along possible trajectories is performed to detect and track targets. Experiment results with real IR image sequence show that the method can greatly improve SNR and target segmentation, and detect and track small moving targets efficiently. It is especially suitable for the cases where a target is partly occluded or missing occasionally.

关 键 词：图像识别 小目标检测 图像预处理 红外图像序列 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置] TP391.41[自动化与计算机技术—控制科学与工程]