Design And Implementation Of A Tobacco Images Grading Algorithm Based On Color Characteristic

China is a large country in tobacco cultivation and consumption, and the tobacco grading is one of the pre-basic works. With the rapid development of the tobacco industry, people put forward higher requirements on tobacco grading. At present, China’s tobacco grading work is mainly carried out by hand according to experience, to judge by eye view and hands touching. This way of grading tobacco has to face some shortcomings such as the low correct classification rate, slow classification efficiency and influence of subjective factors. Based on the color information of the tobacco image this paper design a grading algorithm, and a tobacco automatic grading system based on computer vision technology has been achieved, including transmission of tobacco, lighting and imaging, grading ruling, removed the collection and other functions. This system reached the purposes of enhancing the correct classification rate of tobacco; raising tobacco grading stability; improve the efficiency of tobacco grading purposes. The main research work includes such works:Firstly, this paper reviews relevant research results in this field, analyzing the advantages of the current tobacco grading industry and its limitations. Then according to the problem encountered in the work field of tobacco grading and adapted technique we found a reasonable and workable solution.Secondly, we build an image acquisition system for tobacco grading based on the theory of the solution. Before the task, the performance requirements of tobacco image acquisition system have been proposed, and then we select the relevant hardware to build systems based on economic considerations, feasibility and other aspects. Image acquisition system is not only to achieve tobacco image capturing also enables synchronized transmission, optical illumination, imaging, algorithms judgment, remove, collection and a series of functions.Thirdly, research on tobacco grading algorithm based on Lab color model. In this paper, I propose the three-part grading algorithm, including tobacco image segmentation, tobacco shape pre-grading, and grading algorithm according to tobacco color features. Using Lab model as a metric, I get the distribution of the tobacco color characteristics from known grade tobacco leaves, and according to the color distribution area calculate judgment circular area. Experimental results show that the accuracy of this algorithm improves the correct rate of grading by24%, and the experimental results are stable and meet the expectations.