| Corn is an important food crop in China. The breeding of corn varieties of excellent and seeking for the scientific management is the prerequisite for high yield. Nitrogen nutrition of corn is one of the main agronomic characters of the growth state. Nitrogen nutrition can directly affect the crop leaf color, and indirectly reflect the water requirement of corn, fertilizer and other internal physiological needs. Through the detection of corn nitrogen nutrition to guide precision irrigation, fertilization and other management measures have a great promoting effect on corn yield. So the detection of corn nitrogen nutrition is important to analysis of corn growth and yield prediction work.Nitrogen is one of the most important elements of crop growth and yield and quality. It hasn’t only a certain role in promoting the increase amount of nitrogen on crop photosynthesis and assimilation ability, but also the economic growth and development. It is one of the main factors of crop yield and quality. At present, the appearance and chemical diagnosis method to obtain the diagnosis of crop nitrogen nutrition diagnosis information mainly traditional. The appearance of diagnosis includes three aspects, leaf color diagnosis, symptoms diagnosis and growth vigor diagnosis. When the excess or lack of certain nutrients, people can judge the crop morphology. Chemical diagnosis through a series of complex chemical methods, finally get the specific value of N. Characteristics and patterns of crop symptoms presented, it is due to the physiological functions of nutrients and do not cause.Multi spectral imaging technology with image processing technology and computer technology as the foundation, the image acquisition hardware base, has been more and more widely used in the diagnosis of nitrogen nutrition of crops. Imaging of multi spectral imaging technology by using three dichroic mirrors to link the three CCD, and the white correction influence in many aspects to weeds, soil, weather and so on, to achieve rapid, real-time detection of field crop nutrition information. In agricultural production, nutrient detection and identification of varieties of multi spectral imaging is mainly applied to crops.This paper designs a corn plant nitrogen nutrition diagnosis system, the system is mainly composed of ADC multi spectral camera and Dell notebook computer. For the field acquisition of corn plants image using3×3median filter template de-noising, and then use the Otsu algorithm for image segmentation, the results show that the Otsu algorithm in the segmentation of corn plants at the same time, the maximum extent preserves the spectral information of corn plants, than the other two kinds of algorithm is more suitable for this study.In the corn plant spectral feature value extraction, analysis of multi spectral image11characteristic parameters are obtained, respectively AVSR, AVSG, ANIRNIR, NDVI, GNDVI, RVIR/G, RVIR/NIR, RVINIR/R, r, g, nir. The use of MATLAB software, the jointing stage and Huge bellbottom period from80data sets (11parameters for each group) were analyzed, the BP neural network model was established correspondingly, and test the BP neural network is constructed (60training samples,20groups of test samples). The first randomly selected20groups of training samples, the test on the BP neural network is built, the correct rate of testing network:the jointing stage was95%, the huge bellbottom period of90%. Then the20groups of test samples, test results show that the correct rate of BP neural network, the jointing stage was85%, the huge bellbottom period of80%.Using VC++language implementation of the corn multi spectral image processing system, the system can realize the read image, image pre-processing, image and background segmentation, the results show that the system can accurately provide image processing technology required by the user, which provides the technical guidance of advanced and high precision data support for agriculture expert on corn as matter. |
PDF Full Text Request