Bacterial Angular Sspot Of Cucumber Identification Study Based On The Image Processing

Cucumber is one of China’s main grain crops and it plays an important role in theagriculture production of our country. Pest disease is the biggest obstacle for vegetableproduction. Because of limited knowledge of the pest disease control and prevention,excessive pesticide spraying not only results in over-residue in vegetable products but alsopollutes the environment. Therefore, speedy and accurate identification of vegetable diseases,has great significance on the prevention and control of vegetable diseases.Because of the complex background of cucumber leaf, it is difficult to directly thresholdsegmentation. This paper sets up a spot color detection model for detection of cucumberdisease, and can obtain better results. In this paper, bacterial corner macule of cucumber is theobject of study, and the study is proceeded from the following aspects:1.Image acquisitionCollecting a series of samples of leaf spot disease of the early, middle and late periodsfrom the research field of bacterial corner macule of cucumber in Shanxi AgruculturalUniversity, moreover, several other common cucumber spot images are collected.2. Image preprocessingIt is important to remove noise and improve the image quality through a series ofnecessary operations, because of the impact of the external environment, like light. The firstthing to do is to speed up the processing speed. In order to reduce unnecessary informationand acquire the gray-scale images, we should smoothly gray image. Gray processing and thedenoised image texture enhancement, provide a good condition for the texture featureextraction job.3. Feature extraction and selectionIt is done from the three aspects of color, texture, shape and, extracts15characteristicparameters. Through the optimizing of genetic algorithm, eight eigenvalues are chosen to bethe input of the final recognition system. In order to get better extraction of shape feature, weneed to transform the RGB color space into YCbCr space firstly, and then collect a series ofpoints from lesion and background part and draw the color in the YCbCr space. Through thecolor clustering to establish the clustering model and to achieve the purpose of lesionsegmentation. As for the test results of the lesion, in order to provide better conditions for thesubsequent job, the use of morphology opening operation and closing operation is needed forlesion detection image processing, thus to remove the isolated points and burrs in image,make the interesting parts prominent and lay a good foundation for the remaining extractionand lesion identification.4. Spots recognition The most common BP neural network identification model is used for the diseaserecognition. The disease categories are recognized according to the color and texture features,and then the located periods are recognized based on the shape features of disease.