摘要
探讨了一种红外图像序列中独立运动小目标检测的新方法。通过一种鲁棒的小波多尺度图像配准过程消除主场或背景运动的影响。同时,利用改进的Fisher算法及小波变换对低频的图像进行分割,最终将目标从背景中分离出来,并对配准后分割的图像差分,获得目标运动轨迹。实验表明了该方法的有效性。
A novel system to independently detect small moving object in IR image sequences taken from a moving platform is proposed. Egomotion effects are removed through a waveletbased robust multiscale affine image registration process. The low frequency images are segmented with the improved Fisher algorithm and wavelet transformation, then the object is separated from background. The difference of the segmented image after registration is computed, and the kinematic track of the object is obtained. The experimental result shows that the system is effective.
出处
《红外与激光工程》
EI
CSCD
北大核心
2003年第3期267-270,共4页
Infrared and Laser Engineering
基金
国防科技预研基金项目(5140102021JW0506)
关键词
小波变换
多尺度图像配准
Fisher分割
运动小目标检测
红外图像序列
Wavelet transformation
Multi-scale image registration
Fisher segmentation
Small moving object detection
IR image sequence
