Skip to main content
ARS Home » Plains Area » Lubbock, Texas » Cropping Systems Research Laboratory » Cotton Production and Processing Research » Research » Research Project #434034

Research Project: Color and Lighting Independent Plastic Contamination Detection Via Adaptive Image Processing

Location: Cotton Production and Processing Research

Project Number: 3096-21410-008-55-R
Project Type: Reimbursable Cooperative Agreement

Start Date: Jan 1, 2019
End Date: Dec 31, 2019

Objective:
The global objective of the proposed research is to assess the best method for detecting plastic contamination at the gin-stand feeder apron and in the field. The specific objectives is to develop adaptive translation tracking algorithms that are immune to variations in ambient lighting and able to detect colored plastics that have the same colors as cotton constituents; such as white plastic against a background of white cotton.

Approach:
As all of the modern day image processing techniques such as; DIC and AI analysis are basically simply tools, which are constantly being improved and changing; of critical importance is to establish a rich database of publicly available images such that future researchers can easily gain access to this data-set to try out new and novel techniques as they see fit, develop and adopt from other researchers in parallel research fields of interest. Of particular interest to the cotton world are common plastic contaminants; as such the primary focus of interest will be to place typical source objects into various locations of interest {gin-stand feeder apron and at various locations on cotton plants in cotton fields} with plastic targets sourced from items such as {John Deere module wrap (pink, yellow), gray Walmart sacks, white grocery shopping bags, black plastic mulch, yellow and pink construction ribbons, red and white poly twine (from haying operations)}. Once a sufficiently rich database of images has been collected; various ML, AI and DIC algorithms will be explored for creation of speckle patterns on images collected both in cotton-gins, on gin-stand feeder apron, as well as in cotton-fields (after defoliation). With the primary focus being the application of these techniques to provide a real-time technique that can be used to augment the authors’ cotton gin-stand plastic detection/ejection system for use in dynamic tracking of the cotton velocity as it flows down the gin-stand. By knowing the velocity, it will be possible to develop an automated means to adjust the timing and duration of the pneumatic ejection pulse which in turn dictates the amount of good cotton ejected along with the plastic contaminant. It should be recognized that industry adoption will likely be critically evaluated based on the practicality of the system, so minimization of pneumatic blow-off cycle is felt to be a critically important element in order to achieve widespread adoption of this technology by the industry.