Study On The Skin Defects Of Loquats Based On Hyperspectral Imaging Technology

Skin defects directly affect fruit grading results and reduce the economic value of fruit,and even cause food safety problems.Due to the change of natural environment,some defects will be formed on the surface of loquat in the growth process,the fall in the picking process,and the collision in the transportation process.Skin defects lead to the loquat being easily destroyed during transportation and storage,which causes the risk of other loquats being infected,affecting the selling price.Therefore,the detection of skin defects is one of the urgent problems in the post-harvest grading and processing of loquat.Determining whether a loquat has surface defects can remove unnecessary transportation costs at the source,while avoiding the risk of defective loquats infecting normal loquats.In this study,the effects of spectral data,fusion data,and image data on the detection results of skin defects are compared and analyzed by using loquat.A series of studies were conducted based on the characteristics of three different data,which provided the theoretical basis and technical support for the study of rapid,nondestructive,and high-precision online inspection of fruits.The main research results are as follows:（1）In this paper,hyperspectral imaging technology was used to collect the reflectance,absorbance,and Kubelka–Munk spectra of loquat with skin defects for classification of defects types.Competitive adaptive reweighted sampling（CARS）,successive projections algorithm（SPA）,uninformative variables elimination（UVE）,and Monte Carlo combined with uninformative variables elimination（MCUVE）were used to reduce the dimension of spectral data to obtain the characteristic wavelength.The spectral data and grayscale features were used to establish the spectral model（SPEC）and grayscale features combined with spectral model（MIX）.Extreme learning machine（ELM）,least squares support vector machine（LS-SVM）,and k-nearest neighbors（KNN）algorithm were applied to establish a classification model for skin defects in loquat.Comparing the model classification results of the three spectral parameters combined image features,it was found that the A-CARS-MIX-ELM model had the highest accuracy,with a classification accuracy of 98.18%.The number of selected characteristic spectra was 37,accounting for 21.02%of the total spectral number of the whole band.The hybrid model（Hybrid）was established by ticketing and optimizing the ELM,LS-SVM,and KNN models based on the SPA band selection method.Its classification accuracy was 98.18%,and the number of selected feature spectra was 10 or 11,which was about 5.68%of the total number of full-band spectra.The results show that the optimized Hybrid model can significantly improve the inspection speed while ensuring accuracy,and is more suitable for fast,nondestructive and high-precision online inspection of fruits.（2）The spectral characteristics of a single defect type under different defective degrees were studied.Based on the fusion characteristics of bruised loquat,a qualitative discriminant analysis model was developed for different bruised grade loquats.Firstly,based on the spectral profile characteristics,three different spectral regions,the visible and near-infrared region（Vis–NIR,425-1000 nm）,the visible region（Vis,425-780 nm）,and the near-infrared region（NIR,781–1000 nm）were studied,and the best results were found in the NIR region.Then,based on the selected second PC（PC2）score images,a morphological segmentation method（MSM）was proposed to distinguish bruised loquats from normal loquats.Finally,the multispectral analysis method（MAM）was proposed to detect the bruised degrees of loquats.Different treatment methods are used to reduce economic losses based on different bruised grades.The results show that the proposed MAM can effectively distinguish bruised loquats,and the classification accuracy was 91.3%.The MSM can be used for rapid detection of normal and bruised fruits,and the MAM can be used to classify the degree of bruising of bruised fruits.Consequently,the processed methods are effective and can be used for the rapid and nondestructive detection of the bruised degrees of fruit.（3）Based on the images of defective samples,a method combining band radio image with an improved three-phase level set segmentation method（ITPLSSM）is proposed to achieve high accuracy,rapid,and non-destructive detection of skin defects of loquats.Firstly,principal component cluster analysis combined with contour map were used to analyze the hyperspectral data of three different regions,and the NIR region had the best effect.Based on the weight coefficient curve of the selected PC2 images,the crucial wavelength（782,944.3,and999.3 nm）images was used to establish the multispectral image.Then,the multispectral images were processed again with PCA,and the band ratio image is obtained based on the feature wavelength image.Compare the PC images in the multispectral images and the band ratio images,the two-band ratio image(Q_782/944)is successfully utilized to distinguish skin defects of loquat,it can show a clear contrast between the defective region and normal region.Based on pseudo-color image enhancement,morphological processing,and local clustering criteria,the band ratio image(Q_782/944)has better contrast between defective and normal areas in loquat.Finally,the ITPLSSM was used to segment the processing band ratio image(Q_782/944),with an accuracy of 95.28%.This study indicates that the proposed ITPLSSM method is effective in distinguishing four types of skin defects.Meanwhile,it also effectively segments images with intensity inhomogeneities.Consequently,it provides a theoretical basis for fast,nondestructive,and high precision online detection of fruit.