The Design Of Facilities Of Cucumber Leaf Disease Detection Instrument

In order to promote the development of major projects of "13th Five-Year Plan" key research and development program in the Ningxia Hui Autonomous Region,to realize the nondestructive testing of vegetables,and to promote the development of agriculture.This artical is devoted to the design of non-destructive detection devices for plant leaves.Taking cucumber leaves as the research object,the disease images were captured according to the fluorescence principle of plants.OpenCV was used to deal with the disease image,by this the damage rate of the whole leaf and disease severity were obtained.While using Keras and TensorFlow to complete powdery mildew and downy mildew Disease and the identification of healthy leaves.Using PyQt to design the user operation interface on the raspberry pie,a reasonable detection interface is designed.The user can operate the corresponding button to complete the whole damage rate of the whole cucumber leaf and the detection of the disease type quickly.The main work is as follows:First,According to the research on the chlorophyll fluorescence collection device,have designed a fluorescent collection device with a narrow upper and lower width.The collection device is made of acrylic board.The whole device is economical and practical.According to the characteristics of chlorophyll,selected the blue LED as the excitation source of the detection system in combination with the characteristics of chlorophyll.At the same time,selected white LED as its auxiliary light source.Through the connection with the raspberry pie,we can quickly take photos that meet the requirements.Second,Debugging and control of the device,using the GPIO port to control the small voltage to the large voltage through the relay,thus realizing the coordination of the camera and the power supply,debugging the camera parameters and ensuring the effect of the picture of the blade.Third,Blade treatment,the blade treatment is based on OpenCV.The process of white light leaves are processed to get the whole area of the disease blade,and the spot area is extracted from the leaves of the blue light excited light source,thus the damage ratio and the disease degree of the disease blade are obtained.The image processing of the blade is fast.Complete the entire damage rate test.Fourth,To identify the type of disease,using Keras and Tensorflow to carry out migration and fine-tuning learning on ImageNet data set,build the disease model of blade detection,use the model to distinguish the disease and get the type of leaf disease.Finally.The development of the user interface,specially designed a simple user interface,the interface is clear and simple,the design of the interface is based on the PyQt,can realize the blade shooting,the overall damage rate and disease types by button,quickly an intuitively get the disease condition of the blade.