摘要
图像分割是图像分析的基础,阈值法因简单、快速和稳定而成为图像分割的一种重要方法,在图像处理与识别中广泛应用。针对图像分布具有多样性,本文提出基于随机变量之间关联熵联系的图像分割新算法,该算法通过最大化目标和背景两个概率分布之间的关联熵系数并获得图像分割的最佳阈值。实验结果表明,本文提出的关联熵系数阈值分割方法是可行的。
Image segmentation is the fundamentals of image analysis, and threshodling method is an important approach and is widely applied in image processing and recognizing because of its simplicity, fastness and stability. Considering that the probability distributions of image histogram are various, the new image ~gmentation algorithm based on relative entropy coefficient between random variables is proposed in this paper, this algorithm maximizes the relative entropy coefficient of probability distributions between object and background in order to obtain optimal thresholds of image segmentation. Experimental results show that the thresholding method based on relative entropy coefficient in this paper is feasible.
出处
《西安邮电学院学报》
2007年第1期76-79,共4页
Journal of Xi'an Institute of Posts and Telecommunications
基金
陕西省教育厅项目(编号:06Jk194)
关键词
图像分割
阈值法
关联熵
关联系数
image segmentation
thresholding method
relative entropy
relative coefficient
