Page Banner

United States Department of Agriculture

Agricultural Research Service

Title: White Speck Detection: Image Analysis As a Consistent and Objective Means of Quantifying

Authors
item von Hoven, Terri
item Bel, Patricia
item Negulescu, Ioan - LSU, BATON ROUGE LA
item Collier, Billie - LSU, BATON ROUGE LA

Submitted to: Journal of Cotton Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: March 20, 2000
Publication Date: N/A

Interpretive Summary: The white speck problem results in an estimated $200 million annual loss to the textile industry. Defined as undyeable clusters of undeveloped cotton fibers, white specks appear on the surface of dark dyed fabrics containing these undeveloped cotton fibers. Study of ways to avoid white specks requires a quantitative analysis for them but, to date, no such test was available. This report describes such a test and its application to some fabric samples. The test is based on computerized analysis of a video image. After evaluation of three different image analysis systems, one was selected that consisted of a Microimage Video Systems color camera, an Imaging Technologies frame grabber, a personal computer, second monitor, and Optimas software. Tungsten flood lighting yielded consistent, realistic images. Based on studies of eight plain weave fabrics and thirty-five sateen fabrics, it was found that twenty-four three by four inch images give statistically significant data on the white speck problem

Technical Abstract: White specks are undyeable, undeveloped fiber communities on dyed and finished cotton fabrics that currently are not quantifiable for comparison purposes. In order to remove the subjectivity from dye defect classification, image analysis was evaluated as a means to accomplish white speck detection. Despite the wide variety of imaging software in existence, not all were suited for this application. For this research, two sets of cotton fabrics were evaluated for their white speck content. One study involved eight plain weave fabrics each with visually distinct levels of white speck content and was primarily used to identify the system, software and technique best suitd for white speck quantification. The second study verified the system, software and technique using thirty-five filling face sateen fabrics. The thirty-five fabrics had subtle differences in white speck levels as compared with the eight fabrics. The imaging system was consistent over several dates of testing. The software, hardware and technique were described in detail in this paper as a means to consistently, objectively and repeatedly determine the white speck content of a fabric.

Last Modified: 7/25/2014
Footer Content Back to Top of Page