The Research And Application Of Intelligent Recognition Of Cucumber Disease Based On Mobile Terminal

The new period of agricultural wisdom and modern information techno fogy to make agriculture more closely with the combination to achieve the goal of agricultural science and technology is to realize intelligent information, agricultural production and operation of intelligent and agricultural intelligent life. Things agriculture is a concrete manifestation of the wisdom of agriculture and agricultural machine vision technology letting things more intelligent. Cucumber is a vegetable of the main cash crops, accurate identification of the disease can be done on the forecast and prevention of diseases, the promotion of local economic development plays an important role. With the help of agriculture and machine vision Things theoretical basis of technology to identify common cucumber downy mildew, powdery mildew, leaf spot and anthracnose four kinds of diseases for the purpose of image processing technology as the leading, according to the color characteristics of diseased leaves, Select the appropriate pattern recognition techniques to study the Android smart phone as the representative of the mobile terminal via remote diagnostics to automatically identify cucumber disease, recognition works well, as farmers planted cucumber provides a smart, fast and convenient way to judge anew disease way.Main contents and the results are as follows:(1) With the basic idea of the three-tier system of things and machine vision technology to determine the realization of this project idea. In this paper, the Android smart phone as an image acquisition device, serving as information perception layer, as a client. Network using technology now mature, with good effects 4G data transmission network to ensure reliable transmission of information up and down the line, serving as the network layer; since the smartphone for each image processing library support imperfect, with the computing power compared with PC also there is a big gap, image processing, pattern recognition and the establishment of disease characterized by the library on a PC to achieve, the PC is connected to a remote server, assume the tasks of the application layer, and the result of the recognition process feedback Mobile client.(2) Android smart phone as the main function of the completion of the client is to achieve the image of the collection, storage, cutting and networking and to accept the results of the show. Image collection calls smartphone own high-definition camera, collecting image storage in local disk; call the focal disk disease image, positioning disease outstanding part cutting, cutting can reduce calculation amount of the remote server, but also simplifies the image enhancement and noise image pre-processing, make a clear disease images uploaded to a server. In the process of network upload, use the 4G network, using HTTP protocol to connect to the web server under the environment of distance Tomcat server end, server uses the Struts2 framework technology, good processing the mobile client data transmission to the server. With the combination of Android+Struts2 technology, the image data last lossless transmission is realized. Image through the remote server processing and comparison, and finally to the recognition of the results returned to the mobile terminal display.(3) Build up a complete set of image processing, fast and successful realization of the lesion image segmentation. Select red component grayscale image grayscale image to give the lesion and background contrast grayscale clear, complete image preprocessing; select one-dimensional maximum entropy segmentation method to achieve gray image binarization, and completed the background lesion image segmentation process. Mathematical morphology and image processing algorithms to remove the divided image noise interference, improve and enhance the segmentation results obtained and the original color image size consistent lesion and background separation binary image.(4) The characteristics of we 11-studied lesion, lesion color model was constructed, the establishment of a disease characterized by the parameter library. Use different colors to distinguish different things to achieve recognition in machine vision is the most common way. Binary image after the division of the original gray-scale and color pictures through the field value from processing, binary image of the white area represents the lesion area, normal area on a green background which corresponds to the black area. Using a position corresponding method, using color histogram statistics give the mean lesion area R-cofor, G, B three components. After experimental analysis found that the value of R, G, B three positive linear components will change depending on the intensity of the light changes, the brain perceives color shade each other by the ratio of R, G and B three servings to decide between paper and select R Mean B and G, respectively, the mean average ratio as the selected color feature parameters. Embedded, free installation, portable SQLite database to store the characteristic parameters of the disease. Depending on the time of onset of the disease each color are not the same disease characterization, collection of early and mid-disease multiple image samples blades take the average of each characteristic parameter of disease, and finally the establishment of a disease early every disease signatures medium term.(5) Select the pattern recognition method suitable for the research to identify various diseases. According to the established color characteristic parameter database to compare a variety of pattern recognition methods, choose a template matching the pattern recognition method more categories.30 from each of the four parts of the disease, the use of software identification and ordinary farmers and agricultural workers to identify comparative analysis, the advantage is obvious, the correct recognition rate reached 91.7%.Cucumber Disease intelligent recognition system based on mobile terminals, to achieve real-time, fast, convenient, accurate, non-destructive manner as disease diagnosis cucumber leaves, to solve the lack of errors and mistakes and technicians from the traditional manual visual inspection, saving labor, reduce human subjective factors, the disease did knew early prevention and early treatment, reduce economic losses. Novel methods, application potential, agriculture and development of intelligent precision agriculture has an important significance.