基于Wiener滤波与图像序列相关的小目标检测算法  

Detection Arithmetic for Small Target Based on Wiener Filter and Relativity between Image Sequences

作　　者：李静[1] 赵保军[1] 李栋[1] 

机构地区：[1]北京理工大学电子工程系,北京100081

出　　处：《计算机仿真》2006年第2期90-92,共3页Computer Simulation

摘　　要：传统动态图像跟踪系统在对小目标,特别是点目标的预跟踪阶段,常面临算法较复杂、参数确定困难等问题,在低信噪比条件下不能取得满意的检测效果。该文针对这一问题,提出了一种利用W iener滤波对原始图像滤波,再利用图像序列相关性对滤波后的图像进行“平均背景估计”,从而实现提取目标图像背景,检测动态小目标的新算法。同时对W iener滤波器的参数简化计算以及目标检测阈值的选取进行了详细讨论。该算法参数简单,能在较低的信噪比下取得比较好的滤波效果,仿真结果表明了算法的有效性。Traditionary dynamic image - target track system＇s arithmetic is complicated and the parameters are difficult to be confirmed when tracking small target especially point target, thus usually can＇t get satisfactory result in low SNR. To solve the problem, this article firstly uses Wiener filter for original images, and then introduces an ＂average background estimate＂ method based on the relativity between image sequences for filtered images to pick up the background and finally detect dynamic small targets. The paper also discusses minutely the simple calculation of Wiener filter＇s parameters and the select - value in target detection. The arithmetic has simple parameters and can get better result under lower SNR. The simulation proves the method is effective.

关 键 词：小目标 检测 维纳滤波 序列图像 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]