Image Processing-Based Detection Method For The End Point Of Acid And Zinc Ion Titration

In zinc hydrometallurgy,sulfuric acid concentration and zinc ion concentration of the zinc electrolyte are key parameters for energy consumption control and optimal operation for the electrolysis process.However,the current method of obtaining the acid zinc ion concentration by manual off-line assay results in serious detection and control lag problems.The characteristics of strong corrosion and high concentration of electrolyte,as well as electromagnetic interference and power fre-quency interference of industrial site environment,make it difficult to achieve ideal results with the contact potentiometric titration online detection method.The main research contents are as follows:（1）Aiming at the concentration of sulfuric acid and zinc ions of the zinc electrolyte in zinc hydrometallurgy difficult to detected online,this paper firstly analyzes the functional requirements and design difficulties of the online automatic titration system according to the principle of sulfuric acid and zinc ions titration.Then,a non-contact detection method based on machine vision titration is proposed.Finally,the corresponding titration experiments are designed for the subsequent performance testing and verification of the titration system.（2）In order to solve the problem that the image bubble noise in the ROI during the titration process seriously interferes with the identify-cation of the titration end point,this paper proposes a bubble denoising method for titration images that combines bubble segmentation and bubble hole repair.First,the image is preprocessed by morphological operations,then the bubbles are segmented by the Marker-based Water-shed segmentation method.And finally,the bubble holes are repaired by the expert repair technique.The experimental comparison results show that the no-reference image quality evaluation indexes-the mean,stan-dard deviation and average gradient-which are used to evaluate R,G,and B of the restored image have been significantly improved.The color value and the average gradient of the restored image fluctuates are smaller.The color features are more stable,and the image quality is better.（3）Owing to the time-varying and time-shifting characteristics of the image features in the ROI during the titration process,the titration end point cannot be identified.This paper proposes a similarity method of color frequency histogram with dynamic reference frame to determine the image titration end point.According to the basic principle of image titration and combined with the feature selection and the feature analysis of the titration image,the similarity of the color histogram of the G component is selected to overcome the time-varying of the features,and the time-shifting of the features is overcome by dynamically updating the reference frame.The experimental results show that the proposed method can accurately identify the titration end point,and the average relative standard deviation of each group of results detected by the proposed method is less than 2.5%.The average relative error of the results of each other group is less than 6.7%.The one-sample t test of the results show there are no statistical systematic deviations between the H₂SO₄/Zn²⁺concentration detected by this method and its corresponding true concentration at the 95%confidence level.