ABSTRACT

This paper discusses the identification of small trash objects in cotton using machine vision-based systems. Twenty-one lots of cotton from three growing regions were harvested and processed to test the suitability for high performance spinning. Trash objects were categorized into various size categories based on the equivalent diameter of the objects. The trash distribution from the Cotton Trash Identification System developed at the Southwestern Cotton Ginning Research Laboratory was compared to AFIS and HVI measurements. The machine vision-based systems can evaluate cotton trash, dust, and total counts and were compared to similar AFIS data. The CTIS system developed uses a high resolution camera and can identify objects of smaller size (0.005 mm², 2 pixels) as compared to HVI Trashmeter software (0.045 mm², 1 pixel). Categorization of trash objects in cotton, in real-time, allows for process control and could have a significant impact on the cotton industry.