Detection of Invisible Damage of Kiwi Fruit Based on Hyperspectral Technique

In order to study the method of identifying early hidden damage of kiwifruit, near infrared hyperspectral imaging system in the range of 900–1700 nm is used to acquire the near infrared hyperspectral imaging of sound kiwifruits and damage kiwifruits (in three hours). In this research, kernel-based partial least squares (KPLS) method is used to select the effective bands from 224 hyperspectral bands for reducing data dimension. Then principal component analysis (PCA) is applied to extract features from the effective bands. Finally, the classification result is obtained by the support vector machine (SVM), backpropagation neural network (BPNN) and extreme learning machine (ELM). In the experiment section, the proposed method with band selection based on kernel partial least square is compared with the method without band selection. For 69 sound kiwifruits and 69 invisible damaged kiwi fruits, a total of 138 samples were collected. The best accuracy of band selection based on KPLS method is 98.27%, which is obviously better than the result without band selection. The result shows that the near infrared hyperspectral imaging technique can be used to identify early hidden damage of kiwifruit, and the band selection method based on kernel partial least squares is very helpful to improve the recognition accuracy.

[1]  Haiyan Cen,et al.  Nondestructive detection of chilling injury in cucumber fruit using hyperspectral imaging with feature selection and supervised classification , 2016 .

[2]  D. V. Byrne,et al.  The antioxidative activity of plant extracts in cooked pork patties as evaluated by descriptive sensory profiling and chemical analysis. , 2004, Meat science.

[3]  Stephen Marshall,et al.  Effective Feature Extraction and Data Reduction in Remote Sensing Using Hyperspectral Imaging [Applications Corner] , 2014, IEEE Signal Processing Magazine.

[4]  B. Kowalski,et al.  Partial least-squares regression: a tutorial , 1986 .

[5]  Chih-Jen Lin,et al.  LIBSVM: A library for support vector machines , 2011, TIST.

[6]  Junwei Han,et al.  Novel Folded-PCA for improved feature extraction and data reduction with hyperspectral imaging and SAR in remote sensing , 2014 .

[7]  Zhijing Yang,et al.  Sparse Representation-Based Augmented Multinomial Logistic Extreme Learning Machine With Weighted Composite Features for Spectral–Spatial Classification of Hyperspectral Images , 2017, IEEE Transactions on Geoscience and Remote Sensing.

[8]  Peijun Du,et al.  Novel segmented stacked autoencoder for effective dimensionality reduction and feature extraction in hyperspectral imaging , 2016, Neurocomputing.

[9]  Hadi Seyedarabi,et al.  Potato surface defect detection in machine vision system , 2012 .

[10]  Jon Atli Benediktsson,et al.  Joint bilateral filtering and spectral similarity-based sparse representation: A generic framework for effective feature extraction and data classification in hyperspectral imaging , 2017, Pattern Recognit..

[11]  A. Agarwal,et al.  Efficient Hierarchical-PCA Dimension Reduction for Hyperspectral Imagery , 2007, 2007 IEEE International Symposium on Signal Processing and Information Technology.

[12]  Mingjie Tang,et al.  Detection of Hidden Bruise on Kiwi fruit Using Hyperspectral Imaging and Parallelepiped Classification , 2012 .

[13]  James A. Throop,et al.  Quality evaluation of apples based on surface defects: development of an automated inspection system , 2005 .

[14]  Danilo Monarca,et al.  Near-infrared spectroscopy for detection of hailstorm damage on olive fruit , 2016 .

[15]  Stephen Marshall,et al.  Superpixel based Feature Specific Sparse Representation for Spectral-Spatial Classification of Hyperspectral Images , 2019, Remote. Sens..

[16]  Wenqian Huang,et al.  Detection of early bruises on peaches (Amygdalus persica L.) using hyperspectral imaging coupled with improved watershed segmentation algorithm , 2018 .

[17]  Shiquan Sun,et al.  A Kernel-Based Multivariate Feature Selection Method for Microarray Data Classification , 2014, PloS one.

[18]  Stephen Marshall,et al.  Singular spectrum analysis for improving hyperspectral imaging based beef eating quality evaluation , 2015, Comput. Electron. Agric..

[19]  Wenjiang Huang,et al.  Potential of UV and SWIR hyperspectral imaging for determination of levels of phenolic flavour compounds in peated barley malt. , 2019, Food chemistry.

[20]  Shutao Li,et al.  Novel Two-Dimensional Singular Spectrum Analysis for Effective Feature Extraction and Data Classification in Hyperspectral Imaging , 2015, IEEE Transactions on Geoscience and Remote Sensing.

[21]  Russell M. Mersereau,et al.  On the impact of PCA dimension reduction for hyperspectral detection of difficult targets , 2005, IEEE Geoscience and Remote Sensing Letters.

[22]  Stephen Marshall,et al.  Quantitative Prediction of Beef Quality Using Visible and NIR Spectroscopy with Large Data Samples Under Industry Conditions , 2015 .

[23]  Paul R. Weckler,et al.  Non-destructive quality determination of pecans using soft X-rays , 2007 .

[24]  Stephen Marshall,et al.  MIMR-DGSA: Unsupervised hyperspectral band selection based on information theory and a modified discrete gravitational search algorithm , 2019, Inf. Fusion.