一种新的复杂背景下红外弱小目标检测方法  

New Method for Infrared Small Target Detection under Complex Background

作　　者：王鑫[1] 唐振民[1] 

机构地区：[1]南京理工大学计算机科学与技术学院,南京210094

出　　处：《系统仿真学报》2009年第20期6568-6572,共5页Journal of System Simulation

摘　　要：针对复杂背景下单帧红外图像中的弱小目标检测问题,提出了一种新的基于小波包变换和偏斜度的检测方法。该方法利用小波包对图像进行多尺度分解,解决了高频段分辨率低的问题;并提出了一个基于偏斜度的高斯判别准则,用于对小波包分解系数进行高斯性检验,最终得到了小目标的精确检测。对实测数据进行仿真实验,结果表明该方法能有效检测红外弱小目标。A new method based on the wavelet packet transform and skewness was proposed for infrared small target detection under complex background. The wavelet packets were applied to decompose the image into pyramid subbands at different scales and solve the problem of the higher frequencies having lower resolutions. Then a Gaussian measurement criterion based on skewness was proposed to test the Gaussian nature of the wavelet packet coefficients and finally the exact detection result was obtained. The experiments were performed for some real IR images, and the results prove that the proposed method is effective for small IR target detection.

关 键 词：弱小目标检测 复杂背景 小波包变换 偏斜度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]