虚拟全景影像数据FCM聚类优化仿真  

FCM Clustering Optimization Simulation of Virtual Panoramic Image Data

作　　者：黄艺[1] 龚文辉 HUANG Yi;GONG Wen-hui(Computer Department,Jiangxi Agricultural University,Nanchang Jiangxi 330013,China)

机构地区：[1]江西农业大学计算机系,江西南昌330013

出　　处：《计算机仿真》2022年第11期220-223,239,共5页Computer Simulation

基　　金：江西省教育厅科学技术研究项目(191579)。

摘　　要：传统全景影像数据聚类方法忽略了对聚类目标函数的建立,导致数据聚类结果偏差较大,且效率不高,图像的抗噪性较差。提出基于模糊C均值聚类(FCM)算法的虚拟全景影像数据聚类方法。采用小波过滤变换法对影像去噪和分割。引入FCM算法,结合欧几里得距离、加权平均法构建随机场模型,将目标函数最小化处理,完成对虚拟全景影像的数据聚类。仿真结果表明,所提方法的聚类精度和效率均较高,且图像抗噪性强。Traditionally,some methods ignore constructing clustering objective functions,resulting in large deviation of data clustering results and low efficiency and poor image noise resistance.Therefore,this article presented a method of clustering virtual panoramic image data based on Fuzzy C-Means Clustering(FCM).At first,wavelet filtering transform was used to remove noise and segment images.Secondly,the FCM algorithm was introduced,and then Euclidean distance was combined with weighted average method to construct a random field model,thus minimizing the objective function.Finally,the data clustering of virtual panoramic images was completed.Simulation results show that the proposed method is efficient with high clustering accuracy and strong noise immunity.

关 键 词：小波过滤变换 FCM算法 欧里几得距离 目标函数最小化 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]