基于改进的算术平均算子的阈值分割算法  

Threshold segmentation algorithm based on improved arithmetic average operator

作　　者：兰蓉[1] 史露娜 

机构地区：[1]西安邮电大学通信与信息工程学院,陕西西安710121

出　　处：《西安邮电大学学报》2017年第2期44-52,共9页Journal of Xi’an University of Posts and Telecommunications

基　　金：国家自然科学基金资助项目(61571361);陕西省教育厅科学研究计划资助项目(16JK1709);陕西省科学研究计划资助项目(2014KJXX-72)

摘　　要：针对图像模糊集模型的构造问题,提出一种基于改进的算术平均算子的阈值分割算法。该算法先对指数型限制等价函数进行修正,然后选取5种代表性的隶属度函数,剔除最大和最小的隶属度值,采用算术平均算子对剩余的模糊隶属度值进行集结,将灰度图像转化为一个模糊集合,再以α-型模糊散度为目标函数寻找最佳阈值。仿真结果表明,该算法比其他算法准确率更高,分割效果更好。A threshold segmentation algorithm based on the improved arithmetic average operator is proposed in order to build a fuzzy set model for an image. In this algorithm, based on the selection of 5 kinds of representative membership functions and the modified equivalent function of exponential type, the maximum and minimum membership values are deleted and the arithmetic average operator is used to aggregate the rest of fuzzy membership values to map a grey image to a fuzzy set. Then, an a-type fuzzy divergence is used as the objective function to search for the best threshold value. Simulation results show that the algorithm is more accurate than other algorithms, and the segmentation effect is better.

关 键 词：阈值分割 隶属度函数 模糊集 算术平均算子 

分 类 号：TN911.73[电子电信—通信与信息系统]