Early Detection And SPAD Estimation Method Of Pepper Root Rot

Chili pepper is an extremely important cash crop in China and an indispensable condiment in people’s daily life.By 2022,the national pepper planting areas will reach 30 million ha,accounting for more than 12% of the vegetable planting area.Pepper root rot is a common soilborne disease,it is difficult to detect symptoms which the naked eye in the early stage of the disease,and the traditional control method is to manually detect the density of soil pathogens,but this method is inefficient and costly.Therefore,the early detection of pepper root rot has become an urgent need in pepper production.During the onset of pepper root rot,the chlorophyll content of leaves also changes.The relative chlorophyll content(SPAD)can monitor the growth and health of peppers,so it is important to quickly and accurately estimate the SPAD value of pepper leaves under root rot stress.In recent years,hyperspectral imaging has developed rapidly by virtue of its characteristics of non-destructive testing and has been widely used in crop testing.In this paper,the early diagnosis of pepper root rot based on hyperspectral imaging technology and the estimation method of chlorophyll content of pepper leaves under root rot stress based on spectral index were carried out on the research of pepper leaves No.4 as the research object,and a portable instrument based on the above method was developed to provide new ideas and new approaches for the rapid nondestructive detection of pepper leaf SPAD under the stress of pepper root rot and root rot.The main research contents are as follows:(1)Early detection of pepper root rot based on hyperspectral imaging.The hyperspectral images of pepper leaves in different stages of infection were collected and their reflectivity extracted,the differences in spectral reflectance were analyzed according to the average spectral curves of different stages,13 characteristic bands were screened,and Partial Least SquaresDiscrimination Analysis(PLS-DA)models were established at different stages of infection,and the accuracy of training set and prediction set was 95.1% and 91.1%,respectively.The spectral index with the best detection effect of pepper root rot was screened out through correlation analysis,and the Least Squares Support Vector Machines(LSSVM)model established based on the best spectral index had a prediction set accuracy of 89.7%,which proved that the above two methods had good accuracy for early rapid nondestructive detection of pepper root rot,and provided technical support for the early detection instrument of pepper root rot.(2)Estimation of SPAD value of pepper leaves under root rot stress based on spectral index.On the basis of analyzing the correlation between the predicted spectral index of common chlorophyll and the relative chlorophyll content of pepper leaves,the correlation analysis was carried out by traversing the normalized difference index(NDVI)and simple ratio index(SR)under all band combinations of different spectral transformations(logarithmic transformation,derivative transformation,continuum removal transformation)of pepper leaf reflectivity at400 nm to 1000 nm.The optimal spectral index for predicting the relative content of chlorophyll is screened.Finally,the derivative transformed NDVI(R510,R695)is selected as the optimal spectral index,and a univariate linear regression model is established based on the NDVI to predict the SPAD values of pepper leaves,and the R2 of the training and validation sets are0.857 and 0.823,and the RMSE is 2.97 and 3.05,respectively.Based on the optimal spectral index combined with its corresponding estimation model,the relative distribution of chlorophyll content at the leaf scale is finally visualized,so the SPAD value of pepper leaves under root rot stress can be estimated by using the screened spectral index,which provides technical support for monitoring the growth of pepper.(3)Portable instrument development.The detection instrument adopts Raspberry Pi as the main control processor,and selects and designs the control module,light source drive module,display module and image acquisition module with modular design ideas.Written in Python language,pulse width modulation and image processing technology is used to realize the spectral reflectance information acquisition of pepper leaves,and the model is imported to complete early detection and SPAD value estimation,and the average speed of early detection is 2.3s/time for the accuracy test of the detection instrument,and the accuracy rate is 74.4%;The average speed estimated by the SPAD value was 2.5 s/time,R2 was 0.763,and RMSE was4.58.The results show that the portable instrument can quickly and non-destructively realize the early detection of pepper root rot and the estimation of leaf SPAD value,which provides technical support for the detection instrument of pepper industry.