论文部分内容阅读
Because whether the sugarcane bud was good or not would directly affect the sprouting rate of newly planted sugar cane next year, this paper presented a classification method for sugarcane bud integrity based on Bayes decision in view of the problem that the detection of sugarcane bud integrity could not be realized in present process of sugarcane mechanization planting.On basis of truncating the valuable bud region and extracting five classified characteristics including maximum gray value, this method used computer vision technology to collect the images of sugarcane buds, counted the distribution of image characteristic values in the region of bud, analyzed the distribution curve of this eigenvalues and used percentage between these features that uniform distribution accounted for the entire distribution range and the different percentage of the two kinds of buds in intact and damaged, to determine the characteristics that could be simplified to a uniform distribution and used as the final classification characteristics.According to the full probability formula and Bayes formula, the class conditional probability density function and prior probability of the identified classification characteristics were transformed to posterior probability.The common three sugarcane varieties was selected as the research object, and MATLAB was used to classify the samples to distinguish whether the bud was intact.The result shows that the bud integrity classification accuracy of the three sugarcane varieties is 92.09%、 93.49% and 93.02%, respectively.And the classification accuracy rate of the damaged species reaches 98%, 97% and 96%, respectively.It is provided that this classification method is feasible, and it can achieve the function of integrity classification of sugar cane bud.