Selection of Optimal Wavelength Features for Decay Detection in Citrus Fruit Using the ROC Curve and Neural Networks

Early automatic detection of fungal infections in post-harvest citrus fruits is especially important for the citrus industry because only a few infected fruits can spread the infection to a whole batch during operations such as storage or exportation, thus causing great economic losses. Nowadays, this detection is carried out manually by trained workers illuminating the fruit with dangerous ultraviolet lighting. The use of hyperspectral imaging systems makes it possible to advance in the development of systems capable of carrying out this detection process automatically. However, these systems present the disadvantage of generating a huge amount of data, which must be selected in order to achieve a result that is useful to the sector. This work proposes a methodology to select features in multi-class classification problems using the receiver operating characteristic curve, in order to detect rottenness in citrus fruits by means of hyperspectral images. The classifier used is a multilayer perceptron, trained with a new learning algorithm called extreme learning machine. The results are obtained using images of mandarins with the pixels labelled in five different classes: two kinds of sound skin, two kinds of decay and scars. This method yields a reduced set of features and a classification success rate of around 89%.

[1]  A. Gitelson,et al.  Detection of Red Edge Position and Chlorophyll Content by Reflectance Measurements Near 700 nm , 1996 .

[2]  F. J. Pierce,et al.  The potential of spectral reflectance technique for the detection of Grapevine leafroll-associated virus-3 in two red-berried wine grape cultivars , 2009 .

[3]  Fernando López-García,et al.  Automatic detection of skin defects in citrus fruits using a multivariate image analysis approach , 2010 .

[4]  M. Jimenez-Cuesta,et al.  Determination of a color index for citrus fruit degreening , 1983 .

[5]  José Blasco,et al.  Recognition and classification of external skin damage in citrus fruits using multispectral data and morphological features , 2009 .

[6]  Tom Fawcett,et al.  An introduction to ROC analysis , 2006, Pattern Recognit. Lett..

[7]  Huiqing Liu,et al.  A comparative study on feature selection and classification methods using gene expression profiles and proteomic patterns. , 2002, Genome informatics. International Conference on Genome Informatics.

[8]  Jorge Chanona-Pérez,et al.  Evaluation of Image Analysis Tools for Characterization of Sweet Bread Crumb Structure , 2012, Food and Bioprocess Technology.

[9]  Xianggui Qu,et al.  Multivariate Data Analysis , 2007, Technometrics.

[10]  D. Aboutajdine,et al.  Feature Selection on Supervised Classification Using Wilks Lambda Statistic , 2007, 2007 International Symposium on Computational Intelligence and Intelligent Informatics.

[11]  Lise Korsten,et al.  Integrated control of citrus green and blue molds using Bacillus subtilis in combination with sodium bicarbonate or hot water , 2003 .

[12]  Yiqun Huang,et al.  Applications of Artificial Neural Networks (ANNs) in Food Science , 2007, Critical reviews in food science and nutrition.

[13]  Yibin Ying,et al.  Near-infrared Spectroscopy in detecting Leaf Miner Damage on Tomato Leaf , 2007 .

[14]  José Blasco,et al.  Citrus sorting by identification of the most common defects using multispectral computer vision , 2007 .

[15]  Isabelle Guyon,et al.  An Introduction to Variable and Feature Selection , 2003, J. Mach. Learn. Res..

[16]  Andrew P. Bradley,et al.  The use of the area under the ROC curve in the evaluation of machine learning algorithms , 1997, Pattern Recognit..

[17]  John R. Miller,et al.  Integrated narrow-band vegetation indices for prediction of crop chlorophyll content for application to precision agriculture , 2002 .

[18]  Frank Y. Shih,et al.  Image Processing and Pattern Recognition: Fundamentals and Techniques , 2010 .

[19]  J. Blasco,et al.  Recent Advances and Applications of Hyperspectral Imaging for Fruit and Vegetable Quality Assessment , 2012, Food and Bioprocess Technology.

[20]  Rong-Kuen Chen,et al.  Changes in spectral characteristics of rice canopy infested with brown planthopper and leaffolder , 2007 .

[21]  J. J. Gaffney,et al.  Reflectance Properties of Citrus Fruits , 1973 .

[22]  Ryan M. Rifkin,et al.  In Defense of One-Vs-All Classification , 2004, J. Mach. Learn. Res..

[23]  Bernard Gosselin,et al.  Automatic defect segmentation of ‘Jonagold’ apples on multi-spectral images: A comparative study , 2006 .

[24]  G. Rondeaux,et al.  Optimization of soil-adjusted vegetation indices , 1996 .

[25]  Da-Wen Sun,et al.  Retrospective Shading Correction of Confocal Laser Scanning Microscopy Beef Images for Three-Dimensional Visualization , 2009 .

[26]  J. Gómez-Sanchís,et al.  Advances in Machine Vision Applications for Automatic Inspection and Quality Evaluation of Fruits and Vegetables , 2011 .

[27]  K. S. Banerjee.  Generalized Inverse of Matrices and Its Applications , 1973 .

[28]  J. Aguilera,et al.  Computer Vision and Stereoscopy for Estimating Firmness in the Salmon (Salmon salar) Fillets , 2010 .

[29]  José Blasco,et al.  Multispectral inspection of citrus in real-time using machine vision and digital signal processors , 2002 .

[30]  David C. Slaughter,et al.  Peel Fluorescence as a Means to Identify Freeze-damaged Navel Oranges , 2009 .

[31]  Haruhiko Murase,et al.  Machine vision based quality evaluation of Iyokan orange fruit using neural networks , 2000 .

[32]  Alan M. Lefcourt,et al.  A novel integrated PCA and FLD method on hyperspectral image feature extraction for cucumber chilling damage inspection , 2004 .

[33]  José Blasco,et al.  Detecting rottenness caused by Penicillium genus fungi in citrus fruits using machine learning techniques , 2012, Expert Syst. Appl..

[34]  Lluís Palou,et al.  Control of Postharvest Blue and Green Molds of Oranges by Hot Water, Sodium Carbonate, and Sodium Bicarbonate. , 2001, Plant disease.

[35]  Chee Kheong Siew,et al.  Extreme learning machine: Theory and applications , 2006, Neurocomputing.

[36]  Da-Wen Sun,et al.  Hyperspectral imaging for food quality analysis and control , 2010 .

[37]  G. Camps-Valls,et al.  Hyperspectral system for early detection of rottenness caused by Penicillium digitatum in mandarins , 2008 .

[38]  Antonio M. López,et al.  Feature Selection Based on a New Formulation of the Minimal-Redundancy-Maximal-Relevance Criterion , 2007, IbPRIA.

[39]  Wouter Saeys,et al.  NIR Spectroscopy Applications for Internal and External Quality Analysis of Citrus Fruit—A Review , 2012, Food and Bioprocess Technology.

[40]  David C. Slaughter,et al.  Non-destructive freeze damage detection in oranges using machine vision and ultraviolet fluorescence , 2008 .

[41]  Antonio J. Serrano,et al.  Feature selection using ROC curves on classification problems , 2010, The 2010 International Joint Conference on Neural Networks (IJCNN).

[42]  Noel D.G. White,et al.  Wheat Class Identification Using Thermal Imaging , 2010 .

[43]  Nuria Aleixos,et al.  Erratum to: Advances in Machine Vision Applications for Automatic Inspection and Quality Evaluation of Fruits and Vegetables , 2011 .

[44]  Gustavo Camps-Valls,et al.  Automatic correction of the effects of the light source on spherical objects. An application to the analysis of hyperspectral images of citrus fruits , 2008 .

[45]  Xiuqin Rao,et al.  Detection of common defects on oranges using hyperspectral reflectance imaging , 2011 .

[46]  Amy Tabb,et al.  Identifying Apple Surface Defects Using Principal Components Analysis and Artificial Neural Networks , 2007 .

[47]  Mineichi Kudo,et al.  A Divergence Criterion for Classifier-Independent Feature Selection , 2000, SSPR/SPR.

[48]  Lutz Prechelt,et al.  A quantitative study of experimental evaluations of neural network learning algorithms: Current research practice , 1996, Neural Networks.

[49]  C. Tucker.  Red and photographic infrared linear combinations for monitoring vegetation , 1979 .

[50]  Bim Prasad Shrestha,et al.  Integrating multispectral reflectance and fluorescence imaging for defect detection on apples , 2006 .

[51]  D. Bulanon,et al.  Classification of grapefruit peel diseases using color texture feature analysis , 2009 .

[52]  C. R. Rao,et al.  Generalized Inverse of Matrices and its Applications , 1972 .

[53]  Roberto Battiti,et al.  Using mutual information for selecting features in supervised neural net learning , 1994, IEEE Trans. Neural Networks.

[54]  Franco Pedreschi,et al.  Color of Salmon Fillets By Computer Vision and Sensory Panel , 2010 .

[55]  H. Ramon,et al.  Foliar Disease Detection in the Field Using Optical Sensor Fusion , 2004 .

[56]  Guorong Xuan,et al.  Bhattacharyya distance feature selection , 1996, ICPR.

[57]  J. Qin,et al.  Detection of citrus canker using hyperspectral reflectance imaging with spectral information divergence , 2009 .

[58]  Shiv O. Prasher,et al.  Application of Hyperspectral Technique for Color Classification Avocados Subjected to Different Treatments , 2009, Food and Bioprocess Technology.

[59]  Jon Atli Benediktsson,et al.  Recent Advances in Techniques for Hyperspectral Image Processing , 2009 .

[60]  D. Bulanon,et al.  Spectral reflectance characteristics of citrus canker and other peel conditions of grapefruit. , 2009 .

[61]  Naoshi Kondo,et al.  A double image acquisition system with visible and UV LEDs for citrus fruit , 2008 .