Grouping Method And Realization Of Tobacco Leaves Based On HIS Color Model And Fuzzy Classification

Flue-cured tobacco in China occupies an important role in agricultural production and national economy, with the requirement of increasing the quality of tobacco products, artificial sensory classification and evaluation of tobacco group in China as itself with strong subjectivity and low efficiency, has encountered difficulty to meet the current needs of tobacco grouping, the technology of automatic tobacco grouping therefore became the immediate challenge.The flue-cured tobacco grouping process in the traditional artificial shows high error rates and generally low efficiency. In this thesis, we make research for the realization method of a simulating human sensory tobacco automatic grouping system based on HIS color model and fuzzy classification, in expectation of improving the accuracy and efficiency of the purpose of the tobacco grouping.The main research contents of this thesis are as follows:1. Based on the basic knowledge of tobacco packet, we studied the method of leaf image files reading and displaying, analysis of the tobacco leaf image gray processing, binaryzation and threshold segmentation pretreatment technology, and the use of the small area object removing method in image denoising, in order to achieve the image segmentation and extraction of morphological characteristics in better protection of the tobacco leaf image integrity.2. Compared with the advantages and disadvantages of the commonly used model of color image, we chose more in line with the principle of color based on human visual system:Hue, Intensity and Saturation, the HIS color model for tobacco leaf image color feature parameters acquisition.3. By comparison of the classical fuzzy recognition methods features and scope of use and according to the specific circumstances of the experiment, we proposed a fuzzy tobacco-grouping method based on the principle of maximum membership degree. Through the establishment of membership function based on the tobacco characteristics, we created a tobacco classifier to implement the automatic grouping of tobacco, but also for the establishment. The consummation has laid the foundation of grouping standard feature library.The research indicates a high feasibility of the establishment of a flue-cared tobacco leaves auto-grouping model with digital-image-processing techniques and fuzzy classification methods. In the grouping models, the application of HIS color model as the model of color parameters extraction can make the reliability of grouping increased, and can be applied into a wide range.