浮选泡沫表面动态特征提取方法研究  

Research on the Method of Extracting Dynamic Features of Flotation Froth Surface

作　　者：彭井花 廖一鹏[2] PENG Jinghua;LIAO Yipeng(Concord University College,Fujian Normal University,Fuzhou 350117,China;College of Physics and Information Engineering,Fuzhou University,Fuzhou 350108,China)

机构地区：[1]福建师范大学协和学院,福建福州350117 [2]福州大学物理与信息工程学院,福建福州350108

出　　处：《福建师范大学学报（自然科学版）》2023年第3期94-105,共12页Journal of Fujian Normal University：Natural Science Edition

基　　金：国家自然科学基金项目资助项目(61904031、61601126);福建省自然科学基金资助项目(2019J01224、2019J01087);福建省中青年教师教育科研项目(JAT190973)。

摘　　要：为了解决浮选泡沫图像中光噪点多、相互黏结、无法准确地提取其动态特征的问题,提出了一种浮选泡沫表面动态特征提取的方法.首先,通过分数阶微分最小均值算法、谷底检测算法和形态学处理,得到泡沫边缘轮廓图像,再对这些图像进行三叉点检测,避免了光噪点的影响;其次,用快速视网膜关键点(fast retina keypoint,FREAK)匹配算法对特征点进行匹配,再利用随机抽样一致(random sample consen⁃sus,RANSAC)算法进一步剔除误匹配点;最后,提取出速度特征,并利用特征点对的坐标绘制出速度矢量图和曲线图.实验结果表明,该方法具有更高的抗噪性能,并能够有效改善图像的对比度、减轻泡沫图像光噪点影响、有效剔除误匹配,从而提取出准确的速度特征.本方法的提取准确率为93.3%,该提取准确率较现有一些算法有较大提高,适用于动态变化的浮选工况.In order to solve the problems of excessive optical noise,mutual adhesion,and unable to accurately extract the dynamic characteristics in flotation froth image,this paper proposes a method for extracting dynamic surface features of flotation bubbles.Firstly,the froth contour images were obtained by fractional differential minimum mean algorithm,valley bottom detection algorithm,and morphological processing,and then the three⁃point detection was carried out on these images to avoid the influence of optical noise.Secondly,the fast retina keypoint(FREAK)matching algorithm was used to match the feature points,and then the random sample consensus(RANSAC)algorithm was used to further remove the mismatched points.Finally,the velocity fea⁃tures are extracted,and the velocity vector and curve are plotted based on the coordinate pairs of the feature point.The experimental results show that this method have higher anti⁃noise perform⁃ance,and can effectively improve the contrast of the image,reduce the influence of optical noise of the froth image,effectively eliminate the mismatching,and extract the accurate velocity features.The extraction accuracy of this method is 93.3%,which is higher than that of some existing algo⁃rithms,and is better suited for dynamic flotation conditions.

关 键 词：浮选泡沫 三叉点检测 FREAK匹配 RANSAC算法 速度特征 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]