基于SSIM算法的动态空化图像处理  

Dynamic cavitation image processing based on SSIM algorithm

作　　者：昝晶[1] 马少杰[1] 屠江锋 

机构地区：[1]南京理工大学,江苏南京210094

出　　处：《现代电子技术》2016年第17期23-25,共3页Modern Electronics Technique

基　　金：国家自然科学基金资助项目(51275248)

摘　　要：在空化水洞实验中,需要从大量的数字图像中准确获取空化区域的外形、波动周期等重要信息。首先对比了几种常见的图片边缘提取算法,并选择Canny算子提取空泡结构信息,进而研究了采用SSIM算法对空化图像进行分析的可行性,并通过合理调节结构相似度的比重获得更好的效果。此外,设计了图像的循环批处理程序,该程序通过Matlab编程实现。运算结果表明,采用结构系数为主要特征的相关系数进行分析,更能准确获取空化区域的外形、波动周期等重要信息,更有利于对空化形态进行动态分析。针对一组空化数为0.4的高速摄像图像,该相关系数在0.315~0.375之间变化,空化波动频率约为115 Hz。In the cavitation water tunnel experiment,the shape,fluctuation period and other important information of the cavitation area are needed to correctly acquire in a large number of digital images. Several common image edge extraction algorithms are compared,and the Canny operator is selected to extract the vacuole structure information. The feasibility of using SSIM algorithm to analyze the cavitation image is studied,and a better effect is obtained by adjusting the proportion of structure similarity reasonably. In addition,the cyclic batch program of the images was designed and realized with Matlab programming.The operation results show that the analysis of taking the structure coefficient as the main characteristics of the correlation coefficient can acquire the shape,fluctuation period and other information of the cavitation region more accurately,and it is beneficial to dynamic analysis of the cavitation morphology. For a series high speed camera images with cavitation number of 0.4,the correlation coefficient changes in 0.375~0.315,and the fluctuation frequency of cavitation is about 115 Hz.

关 键 词：水洞实验 空化图像 SSIM算法 结构相似度 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TV131[自动化与计算机技术—计算机科学与技术]