摘要
提出了一种有效的基于递归门限分析的红外目标分割方法。针对传统方法在目标的相对面积较小时背景信息容易误分的问题,将传统分割方法和递归处理结合起来,用于分割红外目标。在分割时,将每次分割得到的背景部分(即暗部分)淘汰掉,而保留分割得到的目标部分(即亮部分)。对得到的目标部分进行再分割,又得到新的目标和背景部分,如此重复下去,直至得到目标为止。对传统的Otsu方法、一维熵方法、二维熵方法的递归分割特性进行了分析比较,并根据目标的先验知识提出一种合理的递归终止准则。试验结果证明,基于递归门限分析的方法是一种行之有效的目标分割方法,分割性能优于传统方法。
An effective infrared object segmentation method based on recursive threshold analysis is proposed. In view of the problem of traditional methods that background information being erroneously divided when object size is relative small, combination of the traditional segmentation method with recursive processing is applied to segmentation of infrared object. The background part (i.e. dark part) obtained in segmentation will be rejected and the object part (i.e. bright part) obtained in segmentation will be retained. Once more segmentation for the obtained object part will bring forth new object and background parts. This process can be repeated until the best segmentation result is produced. Analysis and comparison for recursive segmentation with traditional Otsu method, 1-D entropy method, 2-D entropy method is carried out. A reasonable recursive termination criterion is proposed according to prior knowledge of object. The test results show that the object segmentation method based on recursive threshold analysis is effective. Its segmentation performance is better than the traditional methods.
出处
《光电工程》
CAS
CSCD
北大核心
2004年第10期46-49,共4页
Opto-Electronic Engineering
基金
武器装备预研基金资助项目
关键词
图像分割
红外目标识别
递归分析
Image segmentation
Infrared target recognition
Recursive analysis
