X-ray microtomographic image analysis for identification of cotton contaminants
Technologies currently used for cotton contaminant assessment suffer from some fundamental limitations. These limitations severely restrict the ability of existing technologies to accurately detect and classify contaminants in cotton. Such inaccuracies result in the misassessment of the cotton quality, and have a serious impact on its economic value. The fundamental limitations of existing methods include the inability to detect contaminants under the surface of cotton, the inability to accurately measure shapes and sizes, sample preparation requirements, and poor spatial resolution. These limitations may be easily overcome by the use of x-ray tomographic imaging, which allows for highly accurate imaging of the internal features of an object in a non-destructive fashion. This thesis describes in detail the design of a GUI based interactive cotton contaminant analysis tool. Through the use of an x-ray microtomographic scanner and image processing algorithms, it is shown that x-ray tomographic imaging can provide very accurate information regarding shape, size, and density of cotton contaminants. This information has been analyzed using a fuzzy-logic-based classification scheme to create a highly accurate contaminant analysis tool. Despite its obvious advantages, x-ray imaging does have some drawbacks, principle among which pertains to the time taken to perform the procedure. These drawbacks, along with possible solutions have also been discussed in this thesis. It is our firm belief, however, that if realized in real-time, this procedure will have a definite impact on the cotton cleaning process, and indeed on the economic value of cotton.