一种基于感兴趣区域的红外目标提取方法

Target extraction method based on region of interest

针对复杂背景下的红外图像的小目标提取问题,提出了一种新的红外图像自动分割方法.为了更好地控制计算复杂性及目标提取的准确性,本方法建立在通过自适应 Butterworth 高通滤波器找到的感兴趣区域之上并充分地利用了该区域中背景信息比较简单、目标易被分割的特性.同时,考虑到像素灰度是红处图像区分目标与背景的重要因素,而像素间的相邻度则能较好的防止虚警的产生,结合像素的灰度和相邻度的综合关系提出了一种新的分割原则.实验证明,新算法取得了很好的效果。

In terms of the unportance of the method on target extraction in intrared unages,A new kind of automatic segmentation method is presented.In order to reduce the computation cost and extract target accurately, this method is based on interested region in which background is much simpler and target is easier to be extracted. This interested region through adaptive Butterworth High-Pass filter is obtained.At the same time,considering that pixers value is an important factor to distinguish target and background,and adjacency is useful to prevent false alarms,a novel segmentation rule is proposed which takes full use of the relationship between them.Experimental results show that it is effective.