Design And Implementation Of Monitoring System For Wheat Nutrition Status Using Image Processing Technology

Along with the application of information technology in the developmentof agriculture, nondestructive monitoring technologies based on image processing incrop nutrition increasingly become a hot topic. Nowadays, most monitoring studies onthe nutritional status of the crop throng the image processing technology focused on thefield of cotton, corn, fruits, vegetables, etc. Research on the monitoring of thenutritional status of wheat is less. This paper is based on image processing technology,in order to monitoring the nutritional status of wheat population in field, the pictures of6kinds of wheat varieties under3nitrogen levels were taken by a digital camera andthe nutritional status of wheat population were taken by a SPAD-502at the same time.The images were operated with process of image normalization, image segmentation,image denoising. From the processing, the information of wheat canopy were separatedand12color feature parameters were extracted. And by this, predictive models betweenwheat color features and SPAD were established. Based on the predictive models, thenutritional status of wheat population monitoring system were built using Java Webtechnology.The main contents and results of this paper are as follows:1. The RGB color model was chosen, using pattern recognition method for the imageof wheat canopy population unsupervised segmentation. The operation of imagesegmentation was after the operation of image normalization. Mahalanobis distance incolor space distance metric pattern recognition classifier was constructed, the imagecolor information were classified that achieve the result of wheat canopy in the imagesegmented unsupervised. After image segmentation, image morphological openingoperation and closing operation were taken for denoising. This approach solved theproblem in image segmentation field that crop segment precision in field was not highand the target identification was difficult for a long time. The approach were universaland has practical value.2. From the operation of wheat population image segmentation, the wheat canopy colorinformation was got and from where the12color feature parameters in wheat canopycolor information of RGB color model and normalized rgb color model was calculated.Different estimate models for each variety between the color features and nutrition status of wheat population, and each model passed significant level tests. Each modelreached a higher precision and relative error rates were between4.72%~14%. This laidthe foundation for monitoring the nutrition status of wheat population timely andnondestructively.3. Based on the models of wheat population nutrition status estimation, the remotewheat nutrition monitoring system was designed and implemented that by Java EEtechnology platform, Web-based design three-tier structure, using SSH2lightweightdevelopment framework to MySQL database, Tomcat as Web container. The systembased on B/S model design, including user management, image management,estimation model management, image analysis results management, expert advice andmanagement module. And the system provide the functions of user uploads an image,the user image management, image analysis and the results query, image remotemonitoring and other functions. Through the design and implementation of the remotewheat nutrition status monitoring system, remotely monitor the nutritional status ofwheat for the user become available and provide technical support for farmland timelymonitoring system in future.