Design Of Hydroponic Cucumber Leaf Spot Recognition System Based On Image Processing

With the development of urbanization,hydroponic vegetables have gradually become a part of green life.At present,the recognition of hydroponic cucumber diseases in cultureis still at the stage of combining the existing experience of manual visual inspection and comparison.This method has poor objectivity and low efficiency.Most growers don't necessarily have planting experience,so it is impossible to identify the diseases accurately and timely and make corresponding treatment measures.At present,the cucumber leaf spot recognition system based on image processing is mainly studied in the laboratory.Its applicability is poor in complex environments,and it is difficult to select the combination of disease spot characteristic parameters,so the recognition rate cannot be guaranteed.In this paper,a leaf spot recognition system is designed for cucumber planted in balcony vegetable machine.Image processing technology is used to preprocess cucumber leaf spot images collected under complex background,optimize the combination of characteristic parameters,and improve the accuracy of spot recognition.According to the spot recognition results and disease degree,based on the traditional nutrient solution concentration,the nutrient solution concentration is adjusted to enhance the natural resistance of cucumber to resist the disease invasion.The main work of this paper will be carried out from the following four aspects:(1)Overall scheme design of the system.The overall scheme design of the system includes disease spot recognition module scheme design,nutrient concentration control module scheme design and user management platform construction.The image acquisition scheme was designed to collect the images of diseased cucumber leaves and to design the spot recognition scheme according to the image characteristics.Combining with the architecture of balcony vegetable machine,the type of sensor and actuator is defined,and the nutrient concentration control scheme is designed by referring to the imagerecognition results.Meanwhile,a user management platform is built to facilitate user operation and management.(2)Design and Simulation of cucumber leaf spot image recognition algorithm based on sensitivity optimization feature combination.Taking cucumber anthracnose and powdery mildew as the research object.Firstly,Gauss white noise caused by insufficient light was removed by noise analysis using adaptive threshold wavelet Using ultra-green features,Otsu self-threshold segmentation method combined with morphological operations was used to complete the segmentation of complex background and diseased leaves;The green leaves were removed by color threshold to obtain the spot image for recognition.Secondly,48 original feature parameters of shape feature,color feature and texture feature were extracted.To solve the problem that the combination of feature parameters requires a large amount of work and cannot guarantee the classification accuracy,the GA-BP neural network was used to define the feature sensitivity function of disease spots to optimize the combination of feature parameters and obtain the 8-dimensional optimal combination of characteristic parameters.Improve the search efficiency and reduce the combination of feature dimensions.Considering that spot recognition is mostly small sample classification,the SVM classifier was finally used to classify spot images.The feasibility of the algorithm was verified on the platform of MATLAB 2016 b,and the results showed that the classification accuracy of the optimized feature parameter combination was better than that of the original feature parameter combination.Based on OpenCV and C++language,cucumber leaf spot recognition algorithm is implemented.(3)Design and simulation of nutrient concentration control system based on image recognition results.The relationship between nutrient concentration and detection values EC and PH was analyzed.Referring to the spot recognition results,the fuzzy reasoning system was used to solve the changes of EC and PH values of hydroponic cucumber nutrient solution under the current disease condition.The sum of the set EC and PH values of traditional hydroponic cucumber nutrient solution within the planting days was used as the appropriate EC and PH values for the current disease cucumber.The intelligent control of nutrient solution concentration was completed through PI control.The simulationmodel was established and the feasibility of nutrient concentration control scheme was verified by experiments.(4)Construction of user management platform.Man-machine interface was designed to manage user information,hydroponic cucumber planting information,disease spot identification results and nutrient solution concentration monitoring.The database was developed to store user information,hydroponics cucumber disease spot information,identification results and nutrient solution concentration information,and to trace the life cycle of cucumber.The Experimental and simulation results show that the spot recognition system designed in this paper can segment the spot images of cucumber leaves collected under complex background,and the recognition rate of cucumber anthracnose and powdery mildew is up to 96%,which is 4% higher than that before the combination of characteristic parameters is optimized.Reference image recognition results fine-tuning diseases of cucumber nutrient solution concentration,the experimental results show that the system runs stably,has inhibited or relief of cucumber disease status.The system can also be applied to the disease identification of other crops,providing a reference for the identification of plant diseases in complex environments,and providing technical support for the identification of large-scale crops,promoting the realization of automatic diagnosis of crop diseases.