ASABE Technical Library - Abstract

Abstract. Rapid, reliable detection of cotton foreign matter is a key financial interest of cotton production. While the current industry standard of color grading systems performs well with most types of botanical foreign matter, it often fails to detect white paper and light-colored plastic bag scraps. Pulsed thermography is a form of active thermographic analysis which can detect differences between materials on the basis of several thermal properties. This study explores the use of pulsed thermography to detect and identify cotton foreign matter. Eleven common types of cotton foreign matter, as well as cotton samples, were subjected to pulsed thermographic analysis. Features describing the temperature waveforms and accompanying statistical descriptions of the distributions of these features were derived. Statistical analysis showed that these features and their accompanying statistical features are useful for discriminating between cotton and foreign matter. Classification by linear discriminant analysis using these features achieved 100% accuracy of detection and 91.67% accuracy of identification. These results show that the data produced by pulsed thermographic analysis is suitable and sufficient for detection and identification of cotton foreign matter.