Research And Development Of Defect Detection Device For Typical Diseases Of Korla Fragrant Pear

The disease and defect of Korla fragrant pear fruit directly affect the quality of life,health and safety of consumers.At present,the defect category and degree detection of fragrant pear mainly depends on manual sampling inspection or physical and chemical means.People using their eyes for a long time will cause visual fatigue,and there are deviations in the evaluation standards of different people.The above reasons lead to large identification error and low reliability of Korla fragrant pear fruit disease category and degree.Therefore,there is an urgent need for an accurate,rapid and nondestructive device for detecting fruit diseases and defects of Korla Fragrant Pear.In this study,a portable device for detecting typical diseases of fragrant pear was developed by using visible / near infrared spectroscopy and embedded system.In order to detect the typical diseases of fragrant pear fruit outdoors or on the spot.(1)A device for detecting the type and degree of typical diseases of Korla fragrant pear fruit was developed.The hardware system is mainly composed of control processing unit,spectral data acquisition unit,light source unit,storage unit,display output unit,multi-point detection unit and power supply unit.The control processing unit mainly includes core processor and peripheral interface,which is responsible for data processing and storage;The spectral data acquisition unit mainly includes spectrometer,optical fiber and data transmission line,which is responsible for the acquisition and transmission of spectral signals;The light source unit is composed of a light source and a voltage stabilizing module,which is responsible for providing stable and sufficient incident light;The storage unit mainly includes CH376 module to store the spectral data in rows;The multi-point detection unit includes DC reduction motor and transmission device,which is responsible for driving the rotation of fragrant pear fruit to realize multi-point detection;Touch screen display unit adopts touch screen,which is responsible for user operation input and display output;The power supply unit includes power interface,lithium battery pack and power control circuit,which is responsible for the power supply of different components of the whole device.(2)Through MATLAB,the disease category of Korla fragrant pear fruit and the maximum disease area of Xiangle pear were calculated,and the disease area of Korla fragrant pear fruit in the equatorial area was calculated.The atlas of fragrant pear was studied with time.(3)The classification and degree prediction model of typical fruit diseases of Korla fragrant pear were studied.By comparing four different spectral preprocessing methods,it is found that the best spectral preprocessing method is wavelet transform.Three algorithms were used to classify healthy and different infected grades of fragrant pear fruits.The results show that among the three models,support vector machine(SVM)has the best discrimination performance.For the classification of healthy fragrant pear,infected fragrant pear and infected rotten fragrant pear,the accuracy of SVM classification model based on the optimal wavelength of wavelet transform is 96.67%.SVM algorithm based on wavelet transform is used to classify different penicilliosis grades,and the accuracy is 100%.The SVM algorithm based on wavelet transform is used to classify different rot grades,and the accuracy is 95.83%.