Original papers: Real-time recognition of sick pig cough sounds

  • Authors:

  • V. Exadaktylos;M. Silva;J. -M. Aerts;C. J. Taylor;D. Berckmans

  • Affiliations:

  • Engineering Department, Lancaster University, LA1 4YR Lancaster, United Kingdom;Department of Biosystems, Division M3-BIORES: Measure, Model & Manage Bioresponses, Catholic University of Leuven, Kasteelpark Arenberg 30, 3001 Heverlee, Belgium;Department of Biosystems, Division M3-BIORES: Measure, Model & Manage Bioresponses, Catholic University of Leuven, Kasteelpark Arenberg 30, 3001 Heverlee, Belgium;Engineering Department, Lancaster University, LA1 4YR Lancaster, United Kingdom;Department of Biosystems, Division M3-BIORES: Measure, Model & Manage Bioresponses, Catholic University of Leuven, Kasteelpark Arenberg 30, 3001 Heverlee, Belgium

  • Venue:
  • Computers and Electronics in Agriculture
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper extends existing cough identification methods and proposes a real-time method for identifying sick pig cough sounds. The analysis and classification is based on the frequency domain characteristics of the signal, while an improved procedure to extract the reference is presented. This technique evaluates fuzzy c-means clustering to parts of the training signals and provides a frequency content reference that mirrors the characteristics of sick pig cough. The extraction of the reference is performed in such a way that allows for the identification process to be implemented in real-time applications that would speed up the diagnosis and treatment process and improve animal welfare in pig houses. Preliminary results for the evaluation of the algorithm are based on individual sounds of healthy and sick animals acquired in laboratory conditions. An 85% overall correct classification ratio is achieved with 82% of the sick cough sounds being correctly identified.