Study On Maize Leaf Disease Identification Based On Image Processing

With the rapid development of computer technology, based on the progress of the target shape, size, color, texture and other features in-depth understanding and extraction technology, computer image processing algorithms and pattern recognition techniques to identify and diagnose crop diseases and insect pests is to provide a new research ideas and techniques. Application of Image Recognition in agriculture has made great progress, and become a hot spot of modern agricultural research. This paper is to study the use of computer image processing algorithms and pattern recognition technology to automatically identify crop diseases diagnosis, has been developed to solve the current system can not solve the problem of identifying the disease, but also solve many problems of inadequate staff. Automatic identification of crop diseases for future crop diseases to identify and lay a good theoretical basis.Crop pests seriously affect its yield and quality, this thesis grass crop pest identification is not high degree of automation, to identify problems and timely diagnosis is not taken the computer image processing algorithms and pattern recognition techniques in the diagnosis of maize leaf disease recognition applications to common corn leaf disease as research subjects, and put forward feasible approaches to improve diagnostic accuracy studies provide a theoretical basis for the automatic identification and grass crop pest diagnosis. According to forecast regional stations to detect relevant, current corn leaf diseases are:common rust, leaf spot of corn, a large blotch, leaf blight, curved Aeromonas leaf spot, leaf spot of Aeromonas hydrophila in bent. In this paper, these lesions mainly in corn as the main object of study, collecting corn samples under field conditions disease leaf images, and then according to the characteristics of maize leaf disease on the image of the sample pretreatment, integrated application thresholding, region labeling and so on maize leaf disease image to image segmentation, statistical number of lesions, remove redundant spots, calculated lesion color feature (discussed separately in the conversion method YUV, RGB, HSV color space, such as the color feature extraction algorithm, how and forming a feature vector quantization, and the feature vectors stored in the database, based on the characteristics after reading and searching to find similar images to achieve the whole process, based on this, investigate the effect of reducing the dimension of the vector and the retrieval image block, the introduction of the central region of the color image characteristics weighted color histogram method.) texture features, shape characteristics. Finally retrieval inference diseases. The study made the following progress twofold:first, to achieve the automatic identification of five common corn leaf diseases. Second, comprehensive comparison of identification methods, arrive at a final classification results, improved recognition accuracy and reliability. Expand the range of applications, and the subsequent recognition of corn leaf diseases can also take the form of a computer image processing to solve, saving a lot of manpower and financial resources. Upon examination, the final results retrieved for accurate identification of maize leaf disease has reached more than 80%, the results show that the identification of maize leaf disease is feasible, the paper on the future of maize leaf disease image recognition technology also laid a foundation.