Otsu多阈值快速求解算法  

Quick approach of multi-threshold Otsu method for image segmentation

作　　者：刘艳[1] 赵英良[1] 

机构地区：[1]西安交通大学电子与信息工程学院,西安710049

出　　处：《计算机应用》2011年第12期3363-3365,共3页journal of Computer Applications

摘　　要：最大类间方差(Otsu)方法计算简单,分割效果良好,广泛应用于图像的单阈值分割。为了使Otsu方法能够适应于更加复杂的图像,很多学者对其进行了多阈值的推广,但存在计算量大、效率低的问题。针对此不足,对Otsu方法也进行了多阈值的推广,首先划分直方图区间,然后采用快速二分法求取区间中的阈值以实现Otsu方法的多阈值扩展,使其在保持良好分割效果的基础上大大节省了时间。实验说明了该算法的有效性。Otsu has been widely used because of its simplicity in calculation and good segmentation result.And many scholars have put forward multi-threshold Otsu method to make it more useful in more complex pictures.But this multi-threshold Otsu method is of low efficiency and time consuming.Concerning this shortage,this paper made histogram intervals first,and then the rapid dichotomy was used to seek for the threshold in the interval.The improvement in this paper can maintain good segmentation result and save lots of time.The experiments show that the algorithm is effective.

关 键 词：图像分割 直方图区间 局部阈值 最大类间方差法 多阈值 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]