摘要目的 本文提出一种新颖的基于模糊同质直方图和数据融合技术的彩色图像分割算法。方法 首先计算图像的同质特
征和同质直方图，然后检测出直方图的峰值点对RGB彩色图像各层进行初始分割，最后计算各基色彩色图像的概率分配函
数，使用基于正交和的Dempster-Shafer（DS）理论合并规则进行图像融合，得到最终的彩色分割图像。结果 选用人工合
成和多种医学图像进行仿真实验。定性分析表明基于本文算法的分割图像对比度和清晰度均最优，且图像中细胞边界清晰
完整，细胞数量真实可靠；定量评估结果显示基于本文算法的图像分割敏感度均最高，显著优于现存的基于目标点到原型
成员之间距离的优良模型（Model for Membership Functions，MMFD）和高斯分布假设和直方图阈值（Model Mass Function
Method Based on the Assumption of Gaussian Distribution，MMFAGD）算法，且基于同质直方图优于FCM（Fuzzy C-Means）
和HCM（Hard C-Means）产生的概率分配函数。结论 基于模糊同质直方图的DS证据理论是一种可行的彩色图像分割算
法，不仅能获得优质、稳定、准确的彩色分割图像，而且优越于其他现存的分割算法。
Abstract：Objective This paper presents a novel method of color image segmentation based on fuzzy homogeneity and data fusion
techniques. Methods The general idea of mass function estimation in the Dempster-Shafer (DS) evidence theory of the histogram
was extended to the homogeneity domain. Firstly the fuzzy homogeneity vector was used to determine the fuzzy region in each
primitive color, then, the evidence theory was employed to merge different data sources in order to increase the quality of the
information and to obtain an optimal segmented image. Results Both simulated and clinical datasets were evaluated by different
methods. Qualitative analysis showed that the proposed method, which used both local and global information for mass function
calculation in DS evidence theory, was more accurate than the traditional methods in terms of segmentation quality. Quantitative
evaluation results showed that the proposed method could achieved higher segmentation sensitivity values than HCM and FCM.
Conclusion The experimental results demonstrate the superiority of introducing the fuzzy homogeneity method in evidence theory
for image segmentation.
陆小妍，周啸虎，张子齐. 基于同质直方图和DS证据理论的彩色图像分割研究[J]. 中国医疗设备, 2018, 33(1): 61-64.
LU Xiaoyan, ZHOU Xiaohu, ZHANG Ziqi. Color Image Segmentation Using Fuzzy Homogeneity Method and DS Evidence Theory. China Medical Devices, 2018, 33(1): 61-64.
