Evaluating cotton maturity using fiber cross-sectional images

dc.contributor.advisorXu, Bugao
dc.contributor.committeeMemberChen, Jonathan Y.
dc.creatorOuyang, Wenbin, active 2013
dc.date.accessioned2016-10-25T19:07:12Z
dc.date.accessioned2018-01-22T22:30:52Z
dc.date.available2016-10-25T19:07:12Z
dc.date.available2018-01-22T22:30:52Z
dc.date.issued2016-08
dc.date.submittedAugust 2016
dc.date.updated2016-10-25T19:07:12Z
dc.description.abstractThe cross-section of a cotton fiber provides a directly fiber geometric description. It is known that analysis on a cross-section image will offer a true measure of fiber wall thickness, and derive an accurate cotton fiber maturity evaluation. The fiber image analysis system (FIAS) has been developed for several years. The previous two versions of FIAS were equipped with a traditional microscope with a limited field of view. And the old algorithms were lack of the ability to detect immature fibers correctly, which yielded a systematic bias in the maturity distribution. In this study, images are captured under a new hardware setup with a wide-field of view and a high-resolution camera. A novel descriptor, coupled-contour model (CCM), is introduced to illustrate the relationship between the inner and outer contours of a cotton fiber cross-section. After the detection of the inner and outer contours, triangle-area representation (TAR) is used to describe the shape of the cross-section, and to determine whether the cross-section needs further processing. For those cross-sections analyzed with adhering, self-rolling, scratched, and contaminated characteristics, a cross-section case by case study is required. By analyzed the algorithm efficiency of the randomly picked 7 cottons in 104, it was found the case by case study did occupy about 30% of the whole processing period. This study investigated all the 104 cottons with 15473 fiber cross-sectional images. By introduced more statistic parameters, including mean (Mq), standard deviation (SDq), skewness (Sq), and kurtosis (Kq), a more comprehensive cotton fiber maturity understanding was achieved. According to the maturity distribution, the 104 cottons are distinguishable and divided into 5 classes, i.e. very low, low, moderate, high, and very high from class I to class V, respectively. The comparison made between AFIS and the current FIAS informs that the maturity distributions of AFIS and FIAS are noticeably different. AFIS tends to generate more normal, less skewed and more concentrated maturity distributions but FIAS provides diversified maturity distributions.
dc.description.departmentMaterials Science and Engineering
dc.format.mimetypeapplication/pdf
dc.identifierdoi:10.15781/T2P26Q53M
dc.identifier.urihttp://hdl.handle.net/2152/41977
dc.language.isoen
dc.subjectCotton maturity
dc.subjectImage processing
dc.subjectCross-section
dc.titleEvaluating cotton maturity using fiber cross-sectional images
dc.typeThesis
dc.type.materialtext

Files