Study On Pesticide On-line-mixing Test System Based On Multi-vision Technique

Pesticide spraying is an effective way to prevent from pest attacks and increase agricultural crop yields.However,current pesticide application methods and equipment are inefficient,resulting in wasting pesticides,environmental pollution,and serious ill-health even death to human and animals,and restricting the further development and application of pesticide spraying.The goal of pesticide on-line-mixing is to improve the mixing effect,increase efficiency,and reduce pesticide injury to environment and lives.As people’s standards for environment and food become higher and higher,pesticide on-line-mixing technology has attracted more and more attention.Therefore,pesticide on-line-mixing technology is of grate value to both scientific research and practical application.In this study,the high-speed imaging technology was used to image pesticide on-line-mixing process.It had the advantages of high accuracy,non-contact and good instantaneity,which was very important to mixing effect analysis.Multi-vision technology could capture mixing states in different directions.Using different dimension information could greatly improve the validity and accuracy of the fluid image data acquisition,which provided a new method for mixing effect research of pesticide on-line-mixing.In this article,the stream fields of pesticide on-line-mixing were used as study objects and image acquisition,data processing and mixing effect evaluation were studied by using an multi-vision image acquisition system.Main works are listed as follows:(1)A pesticide on-line-mixing test system based on multi-vision technology was designed and established.The test system consisted of an on-line-mixing system and an image acquisition system.The test system could be used for pesticide on-line-mixing experiments with different flow rates and concentrations,and could simultaneously acquire pesticide on-line-mixing images of two vertical directions.Experiments were carried out under 50 different working conditions,which showed different mixing characteristics in two visions under different working conditions.(2)Pesticide on-line-mixing images were collected based on multi-vision acquisition system and these images were preprocessed before taken into the algorithm.According to the characteristics of the collected pesticide on-line-mixing images,three preprocessing algorithms were chosen.Firstly,a FMM image restoration algorithm based on high correlation of neighbour pixels was used to remove the highlights.Secondly,a calibration method based on standard grids was selected to approximate the distortion coefficients of the images caused by the light deflection.Finally,simple scaling was used to improve the effect of the algorithm and the normalization method of mean subtraction was used to remove the average gray value which was independent of uniformity.(3)A pesticide on-line-mixing uniformity analysis method based on multi-visiontechnology was presented and tested.The uniformity analysis algorithm mainly relied on the idea of local receptive fields and shared weights of convolutional neural network and the definition or description of uniformity by human beings.Firstly,the receptive fields were selected by grid search method.Secondly,two PCA learning models were built for perceptive fields from two different visions respectively.Thirdly,the feature similarity learned from the models was used to calculate the multi-vision pesticide on-line-mixing uniformity.Finally,the proposed algorithm was designed into an analysis software.The results of on-line-mixing uniformity test under 50 different working conditions revealed that the coefficient distribution of different principal components had a certain clustering effect,and the uniformity values of two visions had a certain correlation.With the increase of water flow rate and the mixing concentration,the on-line-mixing effect had a certain improving trend.This results revealed that the proposal method is a promising method to measure and evaluate the pesticide on-line-mixing uniformity based on multi-vision technology,and provided a new way for pesticide on-line-mixing uniformity measurement and evaluation.