摘要
O tsu阈值分割法是目前应用较为广泛的分割方法,其原理简单,计算量小,性能稳定。本文基于其类间方差和类内方差两个评价函数,在连续域内利用微积分原理进行了改进推广,实现了两个评价函数公式的统一,进而在满足文中函数的同时也满足了两个评价函数的要求,给出了快速实现算法。最后,通过实例对比证明,该方法自动选取的阈值合理,极大提高了图像分割的运算效率。
The OTSU threshold segmentation method is one of the image segmentation methods which used extensively for its briefness, small computation and stable performance. Based on its two evaluation functions, between-cluster variance and within-cluster variance, this paper extends the algorithm in mean-field with calculous theory and then realizes the unification of the two evaluation functions, which both meets the requirements of the two functions. And then gives a relative fast realization algorithm. The compared experiment of the paper shows that the threshold gotten from the algorithm automatically is reasonable and the calculation efficiency of the image segmentation has been improved significantly.
出处
《测绘科学》
CSCD
北大核心
2010年第1期26-27,30,共3页
Science of Surveying and Mapping
关键词
OTSU
评价函数
连续域
微积分
Otsu
evaluation function
mean-field
calculous theory
