Research Of Dynamic Feature Extraction Method Of Flotation Froth Based On Image Feature Points

Froth flotation is a common method for extracting the required minerals from ore.The minerals separated from the impurities are adsorbed on the froth in the slurry,so the study of the change of the froth state is extremely important.The previous judgment method was artificial visual observation.When the flotation froth showed stability and not easy to collapse,it proved that the added agent met the separation conditions,but the artificial uncertainty and experience strength required characteristics,and the computer automatic control was gradually applied to the float.Selected.Froth flotation is a common method for extracting the required minerals from ores.The local texture,color,size and other static characteristics of the flotation froth are similar,and it is difficult to be well used for the identification of flotation conditions.Therefore,the dynamic characteristics of the flotation froth are studied.And it can be extracted in real time,which is of great significance for the identification of flotation conditions.In view of the fact that collapse,mergers,and distortions are easy to occur during the movement of the flotation froth,the extraction of the speed characteristics and internal chaos entropy characteristics of the froth flotation can better identify the status of the flotation froth under different industrial mines.The movement of the flotation froth is continuous,and the froth state will continuously change during continuous movement,and the actual state can be better judged under the recognition and matching conditions of continuous frames.Computer automatic control is applied to the dynamic feature extraction of flotation froth.From the perspective of feature point extraction and matching,the dynamic features of flotation froth are obtained by improving the algorithm extraction and matching capabilities.The main research work and innovations of the thesis are as follows:(1)Aiming at the problem of accuracy of feature point extraction of flotation froth,we apply the traditional feature point extraction algorithm SURF algorithm and deep learning algorithm SuperPoint(Self-Supervised Interest Point Detection and Description)to the flotation froth working conditions.Select the research of bubble feature point extraction,improve the use of SURF algorithm and compare the application of deep learning algorithm with this method.Due to the iterative optimization and the superiority of the framework,the deep learning algorithm can extract the feature points of dynamic continuous frames well.Through multi-dimensional experimental comparison with the traditional feature point extraction algorithm,the deep learning algorithm is verified in the extraction of the feature points of the flotation froth image feasibility.(2)Aiming at the characteristics of the continuous movement of the flotation froth and the matching accuracy of the feature points,a feature matching method for flotation froth based on Kalman filtering and RANSAC algorithm is proposed,which is called R-K matching algorithm.Firstly,the feature points are extracted according to the traditional feature point extraction algorithm,and then the sub-block area prediction is performed by the sub-blockdivision of the flotation froth image and the Kalman algorithm,and the initial matching pair obtained by the RANSAC algorithm is used as the initial observation value of the Kalman filter algorithm.Observe the number of eigenvalues in real time and calculate the matching error.When the error is within a certain range,the use of the RANSAC algorithm is stopped.Only the Kalman filter algorithm is used to finally obtain the matching effect map of the flotation froth image.And verify the robustness of the algorithm to confirm its effectiveness and practicability.(3)The simulation results of industrial experimental data show that compared with the existing method,the R-K algorithm can obtain more feature point matching pairs.On the premise of feature point extraction and matching effect improvement,the movement speed and internal entropy of the flotation froth are calculated.The degree of confusion,the stability of the bubble,the disorder and the collapse of the three cases can be well distinguished under the use of the algorithm.Therefore,it can more effectively obtain more accurate dynamic information about the flotation froth under changing conditions,and provides a feasible method for the computer to automatically control and judge the flotation working conditions.