阈值选取是图象处理与分析问题的基础,如何才能正确地找到适当的阈值,是一个古老而重要的问题.目前已有的若干阈值选取方法并不能适应所有的实际图象,常常需根据具体情况自己构造新的阈值选取方法.通过对一种较通用的阈值选取方法构造模式的讨论,提出了一种解决具体问题时构造阈值选取方法的可行思路,并通过一些常用的已有阈值选取方法和一些新的阈值选取方法的导出,对其有效性进行了验证.通过分析,根据实际图象的具体特点可以构造出更具体的(不一定通用)一些阈值选取方法.

Threshold selection is very important for image segmentation. How to get fine threshold is an old, but difficult problem in image processing. We have to make new method for threshold selection in some image analysis because all threshold rule have been put forward for image segmentation can't work well for all images. A pattern about making the method for threshold selection is put forward in this paper. Application in many old methods and some new methods show how to use it to get a new method for threshold selection in practice.