摘要
在对基于相对熵的阈值分割算法(relative entropy-based thresholding)深入研究的基础上,针对目前相应算法(传统算法及最新的改进算法)抗噪性能不强的缺陷,提出了一种新的改进算法。新方法引入了一种新的自适应中值滤波,采用非线性滤波后的中心点及其邻域象素构建共生矩阵,不仅有效地降低了噪声的干扰,而且对不含噪图像和加噪图像都能取得很好的分割结果,在多幅经典图像的分割中取得了满意的效果,实验效果优于对比算法。
On the basis of in-depth study of the relative entropy-based thresholding segmentation algorithm in grey-scale images, contrapose to the deficiencies of the corresponding algorithm at the moment （traditional algorithms and the latest improved algorithm） which do not have a strong anti-noise performance, a new improved algorithm was proposed. The proposed method imports a new adaptive median filter, uses the center point and its 8-adjacent pixels after nonlinear filtering to construct co-occurrence matrix, and get the best threshold via minimizing the relative entropy between an image and its thresholded image. It takes on stronger anti-noise ability, and can achiev very good results of the segmentation for both noise-free images and noisy image. Segmentation in a number of classic images has obtained satisfactory results. Experimental results show that the proposed method performs better than the contrastive methods.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2009年第12期3731-3733,共3页
Journal of System Simulation
关键词
相对熵
阈值向量
图像分割
共生矩阵
relative entropy
threshold vector
image segmentation
co-occurrence matrix
作者简介
胡勇（1972-），男，安徽霍邱人，博士生，研究方向为图像处理、模式识别与机器学习。
