基于LBPV的浮选泡沫图像纹理特征提取  

Flotation froth image texture extraction based on LBPV

作　　者：唐朝晖[1] 朱楚梅[1] 刘金平[1] 

机构地区：[1]中南大学信息科学与工程学院,长沙410083

出　　处：《计算机应用研究》2011年第10期3934-3936,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(61071176)

摘　　要：在浮选生产中,浮选泡沫表面纹理与浮选工况密切相关,直接反映泡沫层的矿化程度(品位高低)。为了给浮选操作提供指导,提出了一种基于LBPV(local binary pattern variance)的泡沫图像纹理特征提取方法。该方法通过融合泡沫图像局部空间结构和对比度来提取泡沫图像纹理特征,然后将LBPV纹理特征应用于浮选工况状态的聚类分析。结果表明,该方法提取的纹理特征能有效反映浮选工况,且能获得更优的浮选泡沫聚类质量。In the flotation process,the surface features of the cleaner froth had close connection with the flotation production condition and reflected the mineralization of foam layer（concentrate grade）.In order to optimize the flotation operation,this paper proposed a method of LBPV to extract the bubble texture features.The features of bubble image were extracted on the combination of local texture spatial structures and contrast features.LBPV texture features used in the cluster analysis of the flotation operation condition state.Clustering results show that the texture feature are efficient in reflecting the bubble states and can obtain better clustering quality of flotation foam.

关 键 词：浮选泡沫图像 纹理 局部二进制模式方差 浮选工况 聚类分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]