Analysis of an Image-Based Fiber Length Measurement Device

No Thumbnail Available

Date

2004-11-07

Journal Title

Series/Report No.

Journal ISSN

Volume Title

Publisher

Abstract

Previous research by Yuksel Ikiz showed that the use digital imaging could provide more accurate and precise fiber length measurements when compared to current methods. This conclusion sparked the research by Stephen Stroupe, which resulted in the development of a system to deliver individualized cotton fibers to a digital camera for imaging. The system individualized fibers from sliver by utilizing a modified comber roll assembly. These fibers were deposited via a chute to an area between two electrodes, where one of the electrodes was charged with 12 kV. The voltage created an electrostatic field that captured the fibers and straightened them. Fibers would then bound between the plates until coming into contact with a non-conductive conveyor. Once in contact with the conveyor, the fibers remained and were passed over a backlighting source, which provided a silhouette of the fibers to a digital camera. The images were analyzed by using the algorithms that Ikiz had previously developed. Stroupe made some initial evaluations of the system's performance and concluded that further analysis would be necessary. It was the goal of this research to begin the analysis and to develop the system on the basis of increasing measurement precision and accuracy. The specific objectives of this research were to improve image quality, evaluate sample selection, improve fiber presentation, determine measurement accuracy and precision, and compare our results to those of other fiber length measurement devices. These objectives were accomplished by using a variety of statistical, programmatic, and mechanical methods. To improve image quality, a pulsing power supply was used in place of a continuous lighting system and was incorporated with a camera calibration routine. To keep the individualizer airflow from blowing long fibers through the electrodes, the comber roll was slowed, which necessitated the removal of the delivery chute. A belt slot was also milled into one electrode for concealing part of the belt to reduce fiber overlap. An additional experiment was conducted to determine the effect of sample size on the measurement repeatability, and to assess the system's ability to distinguish between two different fiber populations. Cut-length rayon fibers were used to assess measurement accuracy. Through the above methods, the image quality was improved, with the contrast between fibers and the background more than doubling their original values. The number of fiber crossovers and entanglements was reduced by an estimated 9 percent with the removal the delivery chute. After cutting the belt slot, fiber overlapping was reduced to 1 mm on average. Although the fibers did not attach to the belt readily after the slot was milled, moving the system to a conditioned environment helped to increase the rate of fiber removal. After all changes, it was determined that our system showed improved accuracy when compared to other systems. However, the accuracy of the results was dependent upon the sample size and day on which the samples were run. The experiment also showed our system was in reasonable agreement with AFIS regarding the difference between the two fiber populations. The cut-length analysis showed our device's ability to measure individual cut-lengths fairly accurately, with a skew toward the lower tail because of the remaining overlap. Despite the existing issues with fiber overlap and broken skeletons, the system was improved through this research and shows promise of becoming a viable method for the measurement of cotton fiber length.

Description

Keywords

SFC, UHM, UQL, Span

Citation

Degree

MS

Discipline

Textile Engineering

Collections