摘要
针对现有熵阈值法在理论上存在的不足和计算复杂性过大的缺陷,提出了一种基于随机变量概率分布之间的关联能量系数的阈值法,方法通过最小化目标和背景两个概率分布之间的关联能量系数推导出了一种选取最佳分割阈值的新算法。实验结果表明,提出的关联能量系数阈值分割方法是可行的,有效的。方法可以弥补现有方法的不足,具有一定的普遍适应性,可以推广到目标和背景符合任意概率分布的图像进行最佳阈值选取。
In view of the shortcomings in its theory and the complexity in its computation of the current thresholding method, a new thresholding method based on relative energy coefficient between probability distributions of random variables is proposed. A new algorithm , which can obtain optimal threshold, is derived by minimizing the relative energy coefficient between two probability distributions of the object and background. The feasibility and effectiveness of this new method are supported by experiments. This method can make up for the deficiency in current methods and also has some universality. It can be widely adopted to obtain the optimal threshold of image segmentation when the object and background follow any probability distributions.
出处
《计算机仿真》
CSCD
北大核心
2009年第5期209-212,共4页
Computer Simulation
关键词
图像分割
阈值法
关联能量
关联系数
Image segmentation
Thresholding method
Relative energy
Relative coefficient
