基于C-means和FCM的侧扫声呐图像分割方法研究  

Side Scan Sonar Image Segmentation Method Based on C-means and FCM

作　　者：田万平 林嘉 TIAN Wanping;LIN Jia(Ship Comprehensive Test Training Base,Naval University of Engineering,Wuhan 430033;Shandong Military Command Data and Information Room,Ji'nan 250000)

机构地区：[1]海军工程大学舰船综合试验训练基地,武汉430033 [2]山东省军区数据信息室,济南250000

出　　处：《舰船电子工程》2021年第11期96-100,177,共6页Ship Electronic Engineering

摘　　要：研究了C-means和FCM两种聚类分割算法对侧扫声呐图像的应用,其中FCM在C-means的基础上引入了隶属度的模糊概念,增加了计算量的同时分割精度有很大提升。同时,对比分析两类分割图像和聚类标准的收敛性曲线。实验结果表明,对C-means、FCM两种聚类算法进行运行速度、分割精度、适用性等方面的比较,发现C-means算法易于实现、运行速度快,但是分割精度不如FCM高,适用于对精确度要求不高的图像分割;而在对比度低、噪声严重的图像区域,C-means算法容易导致误割,FCM算法更合适。This paper studies the application of two clustering segmentation algorithms,which are C-means and FCM,to side-scan sonar images,in which FCM introduces the fuzzy concept of membership degree on the basis of C-means,which increas⁃es the calculation amount and greatly improves the segmentation accuracy.At the same time,the convergence curves of two segmen⁃tation images and clustering criteria are compared and analyzed.The experimental results show that C-means and FCM clustering al⁃gorithms are compared in terms of running speed,segmentation precision and applicability,etc.It is found that C-means algorithm is easy to realize and runs fast,but the segmentation precision is not as high as FCM,which is suitable for image segmentation with low requirements on accuracy.In image areas with low contrast and severe noise,C-means algorithm is easy to lead to miscutting,and FCM algorithm is more suitable.

关 键 词：侧扫声呐 C-MEANS FCM 图像分割 

分 类 号：TB566[交通运输工程—水声工程]