|
| Color characteristics of flotation foam image of Baiyun Obo rare earth mine grade correlation study |
| Received:May 20, 2019 Revised:May 22, 2019 |
| View Full Text View/Add Comment Download reader |
| DOI：doi:10.3969/j.issn.1671-9492.2019.06.018 |
| KeyWord:flotation; bubble; colour |
| Author | Institution |
| Ran Yu |
Inner Mongolia University of Science and Technology |
|
| Hits: 942 |
| Download times: 701 |
| Abstract: |
| In this paper, based on the Baiyun Obo rare earth mine, the correlation between color and grade of flotation foam in flotation process is combined with computer image processing technology, Matlab mathematical analysis software is used to design the algorithm, and the foam image is preprocessed and thresholded. Segmentation, Canny operator edge extraction, bubble color enhancement algorithm, color classification, color pixel area calibration, gray histogram analysis of foam image, calculate and count the overall brightness of the bubble image, as the foam image brightness value in accordance with. In the actual flotation process, the flotation cell has high gas content, bubble overlap, deformation, and the occurrence of bubble merger and rupture. The image acquisition system is designed and set, and the optimal angle of the illumination source is set according to the characteristics of the LED light source. The foam characterization algorithm was optimized to analyze the correlation between the color and grade of the foam image. The results show that the image color feature values and gray scale information are extracted and compared by analyzing and comparing the RGB color value distribution and color classification of the flotation image. The optimized image color compensation algorithm enhances the image effect and improves the color extraction accuracy of the foam image. Through the BP neural network, the foam image is used to represent the eigenvalue color and grade black box model. Through the sample training, the predicted value of the rare earth grade is obtained. |
| Close |
|
|
|