浮选泡沫图像等效尺寸分布特征提取  

Equivalent size distribution feature extraction of flotation froth image

作　　者：卢明[1,2] 桂卫华[1] 彭涛[1] 谢永芳[1] 

机构地区：[1]中南大学信息科学与工程学院,长沙410083 [2]湖南科技大学信息与电气工程学院,湖南湘潭411201

出　　处：《控制与决策》2015年第1期131-136,共6页Control and Decision

基　　金：国家创新研究群体科学基金项目(61321003);国家自然科学基金重点项目(61134006);国家自然科学基金项目(61273169)

摘　　要：利用图像分割方法提取浮选泡沫图像的尺寸分布特征易受光照影响,鲁棒性不强,而利用小波纹理分析方法提取泡沫图像纹理特征则具有多尺度统计特性,对光照鲁棒性较强,但没有形态学意义.针对这一问题,提出一种浮选泡沫图像等效尺寸分布特征提取方法,提取一种新的浮选泡沫图像特征—–等效尺寸分布特征,并将其应用于铜浮选泡沫图像分类识别.实验结果表明,所提取的等效尺寸分布特征可以有效区分3种不同浮选工况所对应的泡沫图像.It is light-sensitive and sick robustness by using the image segmentation method to extract the size distribution of froth image. It is robust to lighting conditions by using wavelet texture analysis to extract froth image texture features because of the multi-scale statistical property of this method, but are not able to reflect froth morphology characteristics directly. Therefore, a method of equivalent size distribution characteristics of flotation froth image based on wavelet multi-scale analysis is proposed. A kind of new flotation froth image features called equivalent size distribution features are extracted, which are used in copper flotation froth image classification. The experimental results show that the extracted equivalent size distribution features can distinguish froth image corresponding to three different working conditions.

关 键 词：泡沫图像 小波分解与重构 最小误差阈值 等效尺寸分布特征 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]