Strips groups detection in images of banknote packages in different lighting and background conditions using an SVM classifier
DOI:
https://doi.org/10.18041/1794-4953/avances.1.1293Keywords:
Binary classification, confusion matrix, histogram concatenation, support vector machines, wavelet transformAbstract
This article shows the results of a binary classification of images with two different lighting and background conditions for a specific case of strips groups detection in banknote packages. The detection is made with a support vector machine classifier trained with features obtained from the images through the application of the wavelet transform and histogram concatenation techniques. For each lighting and background condition a different classifier is trained, the confusion matrix is obtained for each one and then are compared through the recall, specificity, precision, accuracy and Fscore parameters.
Downloads
References
Imprenta Nacional (1991). Constitución políticade Colombia. Santafé de Bogotá.
Colombia, B. d. (2013). Manual de Tesorería.Circular Reglamentaria Externa - DTE -164. Asunto 01: Consignaciones sujetasa verificación y operaciones de cambiode efectivo. Bogotá D.C.
Szeliski, R. (2010). Computer vision: algorithmsand applications. SpringerScience & Business Media.
Lu, D., Weng, Q. (2007). A survey of imageclassification methods and techniquesfor improving classification performance.International journal of Remotesensing, 28, 5, 823-870.
Athinarayanan, S., Srinath, M.,Kavitha, R.(2016). Detection and Classification ofCervical Cancer in Pap Smear Imagesusing EETCM, EEETCM & CFE methodsbased Texture features and VariousClassification Techniques. IJSRSET, 2,5.
Lin, C. (2007). Face detection in complicatedbackgrounds and differentillumination conditions by using YCbCrcolor space and neural network. PatternRecognition Letters, 28, 16, 2190-2200.
Ahmadian, A., Mostafa, A. (2003). An efficienttexture classification algorithmusing Gabor wavelet. Engineering inMedicine and Biology Society. Proceedingsof the 25th Annual InternationalConference of the IEEE, 1, 930-933.
Mallat, S. G. (1989). A Theory of MultiresolutionSignal Decomposition - The WaveletRepresentation. IEEE TRANSACTIONSON PATTERN ANALYSIS AND MACHINEINTELLIGENCE, 2, 7, 674-693.
Chang, T., Kuo, C.-C. (1993). Texture analysisand classification with tree-structuredwavelet transform. IEEE Transactionson image processing, 2, 4, 429-441.
Ghazvini, M., Monadjemi, S. A., Movahhedinia,N., Jamshidi, K. (2009). Defectdetection of tiles using 2D-wavelettransform and statistical features.World Academy of Science, Engineeringand Technology, 49, 901-904.
Manjunath, B. S., Ma, W. (1996). Texturefeatures for browsing and retrieval ofimage data. IEEE Transactions on patternanalysis and machine intelligence,18, 8, 837-842.
Dollár, P., Belongie, S., Perona, P. (2010).The Fastest Pedestrian Detector in theWest. BMVC, 2, 3, 7.
Arlot, S., Celisse, A. (2010). A survey ofcross-validation procedures for modelselection. Statistics surveys, 4, 40-79.
Davis, J., Goadrich, M. (2006). The relationshipbetween Precision-Recall andROC curves. Proceedings of the 23rdinternational conference on Machinelearning, 233-240.
Powers, D. M. (2007). Evaluation: FromPrecision, Recall and F-Factor to ROC,Informedness, Markedness & Correlation,School of Informatics and Engineering,Journal of Machine LearningTechnologies, 2, 1, 37-63.
