Assessing cotton maturity from fiber cross-section measurements

The previous Fiber Image Analysis System (FIAS-I) is not reliable enough to detect fibers, especially for the immature fibers. It yields a systematic bias in the maturity distribution. Furthermore, the maturity distributions are often assumed to be normal without any normality tests in many previous studies, and those distributions are commonly measured by a sole parameter, e.g., the mean maturity value. In fact, those statistical inferences on cotton maturity may not be valid when cotton maturity does not follow a normal distribution. In light of the complexity of maturity distributions, the sole-parameter approach does not appear to be reliable and rational to rank the maturity among different samples. In this thesis, modified algorithms are made in the previous Fiber Image Analysis System (FIAS-I) to improve the number and accuracy of detected cross-sections and reduce the bias on immature fiber. The normality of cotton maturity distributions are analyzed through multiple parameters and patterns of cotton maturity distributions, and the experimental results on the cross section images selected from seven cotton varieties are displayed. Finally, several normality tests are introduced, and the Box-Cox transformation is applied to the maturity distribution, which makes the comparisons among the mean maturity feasible.