Research On Rapid Identification Of Maize Seeds Oriented To Device

Fast, stable and accurate seed varieties pattern recognition is a hot direction of modern agriculture research, and also a problem needed to be solved in agricultural production. Near infrared spectroscopy analysis is low-cost, rapid, nondestructive, meeting the requirements of crop seed varieties identification, and has become the preferred technology of varieties identification.This paper has done research on the various steps of original near infrared qualitative identification process, analyses about the role of each algorithm in identification process. The algorithms are combined together to build up a complete set of near infrared qualitative analysis process. But this method is used in the rapid identification of maize field ineffectively, especially robustness and adaptability of the model are very difficult to satisfy the requirement of application.In order to improve the stability and adaptability of model of Near Infrared Spectra qualitative analysis, using a variety of methods including hidden Markov model for modeling, model transferring with BP neural network. In a conclusion, two methods can improve the robustness of the model in a certain extent, but in the aspect of adaptability, BP neural network model transferring is better than the other method. In general, the two methods are not effective to solve the problem of model transferring.Doing research on the influence of joint modeling combined with different instruments on the characteristics of the model, this method can not only significantly improve adaptability of model, but also can effectively improve the robustness of model. At the sam e time, compared to separate modeling, this method can shorten the modeling time, reduce the workload for modeling, improve the modeling efficiency. Finally, put forward the modeling method of multiple instruments with extending the sampling period to improve the stability and adaptability of model.In order to make this method applied in the special equipment, choose the embedded system. The algorithm transplanted into the embedded system suitable for equipment is beneficial for solving some problems in the application of the algorithm.