Improving Trash Detection Accuracy

Fan You, Mike Mailander, and Fred Sistler


 
ABSTRACT

Lint cotton is classed to determine its grade, which indicates --he market value of the cotton. Trash (non-lint content) is an important factor affecting cotton quality. The objective of this project was to investigate two alternatives to the thresholding method for trash measurement.

In the image analysis, a contrast enhancing (filter) method was introduced to he used with the thresholding method. A new algorithm, topographic labeling, was developed to improve the accuracy of trash detection over that obtained by thresholding. In the final phase of this project, trash detection by conventional thresholding, topographic labeling and contrast enhancing thresholding were compared to trash measurement by the Shirley Analyzer. It was found that both the topographic labeling and the contrast enhancing methods were more accurate than the conventional (HVI) thresholding method.



Reprinted from 1992 Proceedings Beltwide Cotton Conferences pp. 975 - 980
©National Cotton Council, Memphis TN

[Main TOC] | [TOC] | [TOC by Section] | [Search] | [Help]
Previous Page [Previous] [Next] Next Page
 
Document last modified Sunday, Dec 6 1998