Research Of Feature Extraction And Recognition Method Of Three-Dimensional Fluorescence Spectra Of Plant Pesticide Residues

Recently,with the modernization of society and the improvement of human health consciousness,the problem of pesticide residues gradually emerge,which results in more attention have been paid on the detection of pesticide residue from the society.In the process of pesticide residue detection,the "chemical separation" is replaced by a mathematical algorithm,which simplifies the operation procedure,we can continuously observe the pesticide residues and reduce the influences on production and living,which has vital significance to the development of the detection of pesticide residue.Three dimensional spectrogram can reflect the fluorescence feature of samples totally,so we analysis fluorescence curve of selected pesticides by using two dimensional spectrogram and three dimensional spectrogram in this thesis,design a method that can analysis single component resolution qualitatively,through that we found the fluorescent characteristic peak of the pesticide used in experiment.It not only provides standard curves for component resolution,but also provides parameters for selection of hardware devices.We construct the fluorescence spectrogram detection system by choosing the excitation light source,filter and collecting device.We design experiment samples and algorithm routine according to the principle of alternating trilinear decomposition(ATLD),which laid a foundation for collecting data and processing.We used pesticide solution that concentration can be controlled as experiment samples,to distinguish the components in mixed solution of pesticide and distinguish the kind and concentration of the mixed solution,which verified the reliability of ATLD algorithm when be used to distinguish the components in mixed system.Meanwhile,we analyzed the effect of the number of samples and relative fluorescence intensity of component to ATLD algorithm and the proper condition to use ATLD algorithm is proposed.We selected epipremnum aureum leaves which be sprayed single-component and multi-component pesticide solution as experiment samples to collect data and distinguish the component by ATLD algorithm,through the result of distinguish,we verified that ATLD algorithm can estimate the kind and residual amount of pesticides,and it is also verified that the ATLD algorithm is feasibility to detect the pesticide residues.Through the continuous observation of the leaves which have pesticide residues,we analyzed the relationship between the fluorescence intensity of pesticide residues and time and found that with the increase of time,the fluorescence intensity of pesticide residue showed a decreasing trend.The experimental results show that by using the fluorescence spectra detection system with ATLD algorithm,we can distinguish the kind and concentration of components in the mixed solution qualitative;We can distinguish the kind of components in the mixed solution qualitative when use this algorithm to distinguish component of leaf with pesticide residue,it is also can estimate the amount of pesticide residues.