Research On Application Of Image Processing Technology During Curing Process Of Flue-Cured Tobacco

ABSTRACT: To perform a quantitative research on the Change of externalmorphologic features of tobacco leaves during curing, then realize the nondestructivetesting of tobacco leaf’s β-Carotene contents and water contents.450pieces oftobacco images of fresh samples, after curing samples and the key points on theending of the stabilized temperature were selected, image processing technology wasadopted to extraction color characteristics parameters Red, Green, Blue, H, S, B, veinscharacteristics parameters energy, entropy, moment, correlation, and shape featureslength, width, ratio of length and width, horizontal shrinkage, vertical shrinkage,while BP neural networks, genetic algorithm and least squares support vectormachines were used to predict theβ-Carotene contents and water contents during bulkcuring process.The color characteristics parameters R, G, B obviously increased in the period ofyellowing, slowly increased and reached the maximum in the earlier stage offixing-color, and then decreased from later stage of fixing-color to the end offlue-curing, the brightness of tobacco leaves increased first, and then decreased duringcuring, the hue of tobacco leaves decreased significantly during curing, and thesaturation of tobacco leaves changed slowly. The energy and correlation decreased inthe earlier stage of yellowing, increased in the later stage of yellowing, and thendecreased from fixing-color stage to the end of flue-curing.The entropy and momentincreased in the earlier stage of yellowing, decreased in the later stage of yellowing,and then increased from fixing-color stage to the end of flue-curing.Then the color characteristics parameters were used as the input to the BP neuralnetwork with the output of β-Carotene content. The correlation coefficient of theprediction model was0.982, and the average relative error was0.092, which met therequirements of actual bulk curing and provided a theoretical reference for theapplication of image processing technology during the bulk curing.Then the color characteristics parameters and the veins characteristics parameterswere used as the input to the BP neural network, genetic algorithm and least squaressupport vector machines with the output of water content, the mean absolute error and root mean square deviation between the measured data and predicted data from BPneural networks were0.0374,0.0485respectively, and0.0170,0.0200from LS-SVM.The Change of external morphologic features of tobacco leaves during curingcan be precisely quantitative based on image processing technique, and thenondestructive testing of tobacco leaf’s β-Carotene contents and water contents basedon image processing can estimate the β-Carotene contents and water contentaccurately.