Wavelet features selection approach for nondestructive Fusarium corn kernels recognition using spectral data processing

  • Authors:

  • Plamen Daskalov;Tsvetelina Draganova;Violeta Mancheva;Rusin Tsonev

  • Affiliations:

  • Department of "Automatics, Information and Control Engineering", University of Rousse, Rousse, Bulgaria;Department of "Automatics, Information and Control Engineering", University of Rousse, Rousse, Bulgaria;Department of "Automatics, Information and Control Engineering", University of Rousse, Rousse, Bulgaria;Department of "Automatics, Information and Control Engineering", University of Rousse, Rousse, Bulgaria

  • Venue:
  • CSCC'11 Proceedings of the 2nd international conference on Circuits, Systems, Communications & Computers
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

An approach for Fusarium diseased corn kernels recognition based on wavelet continuous transformation and probabilistic neural network (PNN) classification is presented in the paper. The near infrared diffuse reflectance characteristics are used as base for features extraction. The spectral data are fitted using continuous wavelet mexican hat transformations and their parameters are used for classification features. The range of the wavelet parameters "scale" and "time" are defined for seven most popular corn kernels variety in Bulgaria. A PNN classifier is created using wavelet coefficient Ca,b. The range of the PNN classifier smoothing parameter s is determined experimentally. The classification accuracy is in the range of 75 to 100%.