Research On The Diagnosis Method Of Crop Canopy Diseases Based On Thermal Infrared Image Processing Technology

The real-time diagnosis of crop diseases and the application of pesticide control are of great significance to the high-quality and high-yield crops in the natural environment.Thermal infrared images of crop canopy can characterize its reproductive growth information.Compared with traditional single-point measurement technology,it has scale advantages and is one of the important means to detect early plant diseases in real time.At present,the application of crop canopy thermal infrared image processing technology to build a rapid identification method of crop diseases that is difficult to describe with accurate mathematical model has become a hot and difficult issue in the research of intelligent agriculture.And this research used crop canopy dual-source images as the research object,deeply integrated plant pathology knowledge and intelligent information processing technology,extracted the canopy visible light reference image,identified the target area of the canopy thermal infrared image,and established Taking the northeast red bean rust as an example,the research on the diagnosis method of crop canopy diseases based on thermal infrared image processing technology was completed.The main research contents are as follows:(1)The visible light image recognition model of crop canopy based on fuzzy neural network has been established.Canopy visible light image of the crops was used as the object.First of all,the adaptive characteristics of the five-layer linear normalized fuzzy neural network were used to select the Gaussian membership function to automatically calculate the inference rules for canopy visible light image recognition and effectively identify visible light Canopy area in the image.Of the image canopy,through the calculation of the image canopy effective identification index,entropy value,histogram and so on.The canopy segmentation quality of the visible light image was quantitatively evaluated.The average accuracy of this algorithm to effectively identify healthy canopies and affected areas was 96.54% and 96.86%.The average pixel information entropy of visible light canopy images was 4.2425 and 4.7137,which was in line with the pixel information entropy of canopy images obtained by standard algorithms.The difference was only 0.4712,which provides an effective reference image for obtaining thermal infrared target images of crops.(2)A thermal infrared target image recognition model based on the affine transformation algorithm has been established.The canopy area of the visible light image was used as the reference image,the registration parameters of the reference image and the original thermal infrared image were first calculated by affine transformation to identify the canopy target area of the thermal infrared image.For the thermal infrared image of crops with the initial temperature range of 14.04 ~ 20.00 ?,the maximum temperature difference of the target image was 4.99 ?,and the average temperature value of the original image was reduced from 17.2110 ? to16.4637 ?,based on thermal infrared image processing.The technology of crop canopy recognition,and then through the information entropy to evaluate the crop canopy thermal infrared image recognition method,the average entropy values of thermal infrared canopy images of healthy and diseased samples are 4.2282 and 4.7881,respectively,for the extraction of crop canopy thermal infrared images The canopy temperature characteristics provide a real and reliable data source.(3)A crop canopy disease diagnosis model based on thermal infrared image features has been established.Based on the crop canopy temperature information,the maximum,minimum,average,mean,variance,standard deviation,entropy,temperature variation coefficient,temperature normalization value,and temperature frequency of different frequency bands of the canopy temperature were calculated.Based on the principal component analysis method and the8-dimensional temperature characteristic parameters,the intelligent neural network,multiple linear,BP neural network regression were used to establish an intelligent rapid diagnosis model for crop canopy diseases.30 groups of samples were used to verify the accuracy of early disease detection and prediction models.The accuracy rates were 80.00%,86.67%,96.67%,which can provide technical support and reference for the application of thermal infrared image technology to the diagnosis of field crop diseases.Based on the visible and thermal infrared images of the crop canopy Feng from different growth periods,this study identified the canopy targets of the thermal infrared image based on the extracted visible light reference images,optimized the temperature characteristic parameters of the multi-dimensional disease area,and dug deep The non-linear mapping relationship between canopy disease symptoms and temperature characteristics finally established a methodfor rapid diagnosis of crop diseases based on thermal infrared image processing technology.The research results provided technical support for early automatic rapid diagnosis and accurate prediction of crop diseases.