基于多级滤波的复杂背景下多尺度小目标检测  

Multi-scale small targets detection in clutter based on multi-level filter

作　　者：王岳环[1] 程胜莲[1] 周晓玮[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国家教育部重点实验室,湖北武汉430074

出　　处：《红外与激光工程》2006年第3期362-366,377,共6页Infrared and Laser Engineering

基　　金：国家自然科学基金重点项目(60135020);国防重点预研基金项目(51401020203JW0501)

摘　　要：针对精确制导系统中目标尺度的不确定性带来的检测问题,在分析原始多级滤波器结构与传递函数频率特性的基础上,适当改进了多级滤波的结构,导出多个滤波结果,并提出了基于融合多级输出滤波结果的多尺度小目标检测算法。实验结果表明,该算法可以在没有目标尺度先验信息的情况下自适应地检测出多个小目标。In precise guiding system, the sizes of the target image varied because of the relative movement between the target and the imaging system. To solve the detection problems from uncertainty of object sizes, an improved structure of multi-scale filter was proposed based on analysis of original structure and frequency characteristics of transfer function. An algorithm of multi-scale small targets detection was also developed based on the multi-level outputs of the new filter. Experimental results demonstrate that the algorithm proposed in this paper can detect multi-scale small targets in situation without any prior information about the sizes of the target in advance.

关 键 词：多尺度小目标 目标检测 复杂背景 多级滤波 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]