Selection Based On Color And Texture Characteristics Of Mineral Flotation Bubble Classification

Froth flotation is a widely used process to extract certain minerals from ore, which involves extremely complex physical and chemical processes. At present, the control of mineral flotation process is often based on the visual appearance of the froth state, and to a large extent depends on the experience and ability of a human operator. These types of process are consequently often controlled sub-optimally owing to high personnel turnover, inaccuracy or unreliability of manual control. Consequently, the research of the froth image recognition and applyication to practice is of great significance for optimizing flotation process.Several characteristic(color,texture) from image characteristcses is selected to describe the concentrate froth image and the image processing techniques to automatically measure the color and the texture parameters of froth bubbles is studied and the relationship between those feature parameters and metallurgical process datum is also analyzed.Then the method of recognizing all sorts of concentrate images belonging to different types is studied based on the SVM.First, the color and the texture parameters of froth image are extracted to describe the concentrate froth image based on the analyses of the characteristics of froth image and the concentrate bubble is difficult to segment for its tiny. Color analysis is performed on the whole surface of the froth images considering different color reference system.The PCA was used to extract the froth color texture. Experiment analysis shows this parameter not only save the most information of the image color,but also can reflect the grade and recovery of mineral efficiently.Then, it is hard to establish the link between the texture and technics parameters. considering the limitation of the texture characterization approach in the bubble image, an effective texture characterization method is proposed for accurate description based on the fuzzy texture spectrum.The relativity was analyzed between the extracted texture and the concentrate of mineral. The results prove their effectiveness. Finally, Considering the problems of inaccurate classification for small simple,nonlinear and high dimension in classic method,a support vector machine method based on statistical learning theory was proposed. Using the froth color and texture as the input,the paper established the image state recognition model and realized the classification and recognition of the concentrated froth image. Experimental results prove that this model has good performance of recognition which has established the foundation for flotation process adding medicament control.