Research On Industrial Application Of Feature Extraction Algorithm For Flotation Foam Image

Flotation is the most widely used mineral processing technology,the visual characteristics of the flotation foam are often directly reflected in the flotation conditions.For a long time,the control of the flotation process has been based on the subjective visual judgment of workers that has lots of trouble.In recent years,advances in digital technology and automated control have made the extraction of feature-based flotation foams based on computer vision popular.This article analyzes the difficulty of the feature extraction algorithm from characteristics of Flotation on-site foam image,and put forward the corresponding algorithm improvement.The main work is as follows:(1)Analysis of flotation process: Summarize the correlation between the characteristics of flotation foams and operating parameters,and point out the types of features that need to be extracted.Analyze the difficulty of image processing algorithmic based on characteristics of Flotation on-site foam image,and propose the idea of extracting bubbles.(2)Due to the oversegmentation and edge shifting caused by when the watershed algorithm,an optimization idea is proposed.Firstly,to achieve de-noised and enhanced by image pre-processing,which avoids the over-segmentation caused by noise and enhances the fuzzy edge details.And use FCM clustering algorithm for image tagging,which makes the algorithm adaptive.In addition,optimize the mechanism of watershed function,and precise partitioning was achieved through algorithm optimization,which increases reliability of foam static(3)For complex additional conditions in the foam flow rate,to use ORB feature point matching algorithm to estimate foam velocity,and realize the real-time performance of flow rate estimation.Analyze the existing problems of ORB algorithm and improve the algorithm that uses the merging block matching method and feature point matching method are used to integrate the advantages of the matching algorithm.Thus,the accuracy of the foam flow rate estimated is improved,to obtain more accurate dynamics features.(4)In the Jinchuan nickel ore production line,a hardware platform for the flotation foam feature analysis system was adopted,and the software functional structure design was proposed based on functional requirements analysis.On the basis of realizing on-site video monitoring and feature data extraction in flotation,feature data forms are enriched,and data storage and interaction functions are added.In addition,based on the user's usage angle,the user-friendliness of the interface is improved,and the system operation is automated and user-friendly.