The Algorithm Research And Software Design Of Heavy Metals Detection System Based On Nano-Gold Membranes

The heavy metal pollution has been becoming more and more serious with the industrial development and economic growth of China.It has obvious significance for the prevention and control of heavy metal pollution by detecting the content of heavy metals timely.There are some insufficient when heavy metal detection by using the mass spectrometry,spectrum,bioanalysis,etc.Such as time-consuming,expensive,complex,etc.But nano-gold based membrane has some advantages when detecting the heavy metals,such as stable,accurate,simple to operate,etc.So,heavy metal detection system research based on nano-gold membrane has higher practical value,it‘s a supplement and development of the heavy metal detection techniques.The algorithm research and software design play the essential roles of the detection system,there are some important theoretical value and application prospect in them.This thesis presents the deep research of the algorithm and detailed design of the software for the detection system,solves the essential problems of the system.To address the inefficiency and inflexible of existing methods to extract the sensor‘s color features manually,a two-stages image processing algorithm based on HSI color space and seeded region growing(SRG)is proposed to achieve automatic,precise and high-throughput features extraction of nano-gold membranes.On the basis of the color features,uses support vector regression(SVR)model to achieve accurate concentration recognition.Finally,constructs the Linux system environment on ARM11 board,uses Qt-Creator coding to achieve system software design.Specifically,the main contents are as follows:(1)Based on the features of nano-gold membrane,design a two-stages image processing algorithm based on HSI color space and SRG for membranes array.The first stage transforms array image from RGB space to HSI space,completes rough segmentation in HSI color space,completes image filtering based on mathematical morphology algorithm,completes gridding based on projection algorithm.On the basis of the first stage,the second stage completes the precise segmentation by proposing a modified SRG algorithm,and saves the color features in color map finally.(2)This thesis uses SVR model to process the color features to achieve concentration recognition of heavy metals.Specifically,chooses the appropriate training set and test set,uses 6-fold cross validation method to choose the optimal parameters of SVR model,uses RBF kernel,Sigmod kernel,polynomial kernel to train the soft ?-SVR model and ?-SVR model respectively.Then uses the recognition results of soft ?-SVR model and ?-SVR model to compare with the results of polynomial nonlinear regression and BP neural network.Finally,?-SVR model is chosen to be the recognition algorithm of this thesis.(3)According to the software development processes,analyzes the software requirements,designs the overall design of the software,then achieves the functional design of every module.Constructs the Linux system on Tiny6410 board,uses Qt coding to achieve the user interface design and functional modules design.Tests the software finally.The results of algorithm research and software design demonstrate that the proposed image processing algorithm can extract the color features of each reaction spot in nano-gold membranes array precisely,stably and adaptively.The proposed concentration recognition algorithm has higher recognition precision.And the function of the software is normal,the software can be used in real application scenarios.