白云鄂博稀土矿浮选泡沫图像颜色特征与品位相关性研究

Study on the Correlation Between Color Characteristics of Flotation Foam Image and Grade of Rare Earth Ore in Bayan Obo

在稀土矿物实际浮选中,泡沫颜色特征与稀土品位关系密切。针对白云鄂博稀土矿浮选过程中浮选槽中含气率高、气泡重叠、变形,以及不断发生气泡兼并与破裂的情况,设计建立图像采集系统,并针对LED光源特点,设定照射光源最佳角度,进行图像采集。对采集图像进行进一步颜色特征提取,对泡沫图像颜色与品位之间的相关性进行分析研究。根据浮选过程中浮选泡沫表征颜色与品位的相关性,结合计算机图像处理技术,使用Matlab数学分析软件,对泡沫图像进行预处理并且进一步对泡沫图像进行边缘提取,对泡沫色彩进行色彩效果增强处理,将颜色分类量化,并进行色彩分类统计。通过对泡沫图像灰度直方图分析,计算并统计其整体亮度情况,作为泡沫图像亮度值定量依据。结果表明:通过分析浮选图像RGB颜色值分布、颜色分级分类量化提取图像颜色特征值以及灰度信息,对泡沫颜色特征有一定代表性,并且提高泡沫图像颜色提取精度。通过BP神经网络,输入泡沫图像特征值颜色与品位信息并建立黑箱模型,通过样本训练,得到稀土品位预测值。

In the actual flotation process of rare earth ore,foam color characteristics and rare-earth grade is closely related.Aiming at the high air content in the flotation tank,bubble overlap and deformation,as well as the continuous merger and rupture of bubbles during the flotation of rare earth ore in Bayan obo,an image acquisition system was designed and established,and according to the characteristics of LED light source,the best angle of irradiation light source was set for image acquisition.Further,the color features of the collected images were extracted,and the correlation between color and grade of the foam images was analyzed and studied.For flotation foam flotation process characterization of the correlation of color and taste,in combination with computer image processing technology,the use of Matlab mathematical analysis software,the bubble image preprocessing and further carried out on the foam image edge extraction,to color effects to enhance the foam color,quantification,color classification and color classification statistics.By analyzing the gray histogram of the foam image,the overall brightness of the foam image is calculated and counted.The results show that:by analyzing the distribution of the RGB color value of the flotation image,extracting the color feature value and gray information of the image by colorclassification and quantification,the foam color feature is representative to some extent,and the color extraction accuracy of the foam image is improved.BP neural network was used to input the color and grade information of the characteristic values of the foam image,and a black-box model was established.The predicted value of rare earth grade was obtained through sample training.