ABSTRACT
Aims Over a long period, cottons have been mostly harvested in either manual or machinery way in China. The former is highly subjective, which is not accurate for grading, while the latter is not able to grade at all, which may leads to the decrease in the yield of high quality cotton. Therefore, it is necessary to come up with an approach to grade the preharvest cottons. At present, although much research has been done on ginned cottons for grading outside and in by HVI equipment, little work has been done on unginned cottons as field preharvest cottons for grading outside. In order to assess the quality of preharvest cottons with bracteoles objectively, this thesis presented the results of experiments in which ranking classifiers was designed by employing machine vision and pattern recognition to sample the preharvest cottons with bracteoles and then grade them based on their sizes and colors. The application of the classifiers in cottons harvest could lesser the burden of grader, improve the accuracy of the grading, accordingly, increase the yield of harvesting high quality cottons.
