Barley Diseases Classification And Identification Research Based On Machine Vision In Gansu Province

Barley is an important raw materials for beer production, also it is an importantfood and livestock feed in chinese Gan-Qing-Zang area. Along with the developmentof science and technology, especially the development of digital agriculture, precisionagriculture,machine vision is widely used in any branch of deficiency symptomsdiagnosis,Seed Testing,the quality of agricultural products and classification, haveachieved many achievements.Barley are susceptible to pests and diseases in the growth period with othercrops,which will reduce the yield, restrict the development of agricultural economy.The research on the basis of analysis the developing status of the crop pest detectionand recognition at china and abroad,proposes a new method for detecting the differentwith the traditional method. Digital processing of barley pest image, and thencombined with pattern recognition to achieve the Classification. The results indicatedthat the classifier can identify crop pest category,and had a good result..It has apositive effect to guide the barley plant protection work of agriculture. Application ofmachine vision technology has a great significance in accurately and quickly detectand identify the disease of barley, timely prevention, reduction of barley yield loss,and improve the quality of barley has important practical significance.The studies were conducted in the following aspects:Image acquisition and preprocessing is the key step for the subsequent imagedata operation, which direct impact on the effect of image processing, and furtheraffect the accuracy of barley disease recognition. Images were collected in the naturalbackground, and uses the B channel grayscale, then use the median filtering,Shuangfeng method and watershed algorithm method for image processing.For feature extracted of image color,respectively R, G, B three channel to extractcolor moment features,including first moment and second moment of threechannels.Indicated by the test data, this feature has a good discrimination for barley normal leaf and disease leaf.In the aspect of texture feature extraction, Choose local binary pattern. Throughthe calculation of LBP values of image data for the extraction of texture feature.Paperselects the color and texture feature for an comprehensive characteristic used toidentify the barley diseases.Use support vector machine (SVM) to construct the classifier, the classifier forclassification has an better classification effect. For barley overall disease recognitionresult up to84%.