摘要
对Otsu准则的内在缺陷进行了改进,提出了一种基于直方图的分割方法.通过对直方图的预处理及轮廓追踪,可以很好地确定其峰谷结构,从而找到理想的分割门限.这种分割方法对红外图像具有很强的针对性.经实验证明,只要图像直方图中存在着多波峰结构并且存在一个理想的分割门限。
Aimed at the inherent defection in Otsu Law, an improved method of segmentation is proposed based on histogram. After previous treatment and tracing the frame, it is easy to identify the peaks and villages of the histogram. The threshold of segmentation can be found accurately. The method is especially useful for the segmentation of infrared images. According to the experiments, the effect is good if there are waves and an ideal threshold exists in the histogram.
出处
《华中理工大学学报》
CSCD
北大核心
1999年第1期84-86,共3页
Journal of Huazhong University of Science and Technology