Hyperspectral Detection For Wheat Stripe Rust And Monitoring Method Of Urediospores In Field Air

Wheat stripe（yellow）rust,which is caused by Puccinia striiformis f.sp.tritici（Pst）,is a prevalent and distributed wheat disease across the world,particularly in cool and moist regions.It is one of the most important and devastating airborne wheat diseases in China,and has caused severe yield reduction,resulting in significant economic loss.Wheat stripe rust is an important target for prevention and control in China,which has invested a large amount of manpower and material resources in the investigation and monitoring of wheat stripe rust.However,due to the lack of effective techniques for early monitoring and forecasting of the disease,wheat stripe rust is easy to epidemic and outbreak,which has brought great losses to wheat production.Stripe rust of wheat is a kind of airborne multi-cycle fungal disease,and the source number of urediniospores is the main factor affecting its occurrence and epidemic.Traditional wheat stripe rust disease investigation and stripe rust pathogen spore monitoring method have the disadvantages of large labor,low efficiency,and reduced accuracy with working time,which makes it difficult to grasp the real-time and dynamic changes of large-scale farmland diseases and fungal spores.In order to quickly and accurately evaluate the degree of wheat stripe rust disease and to achieve the remote real-time quantitative monitoring of urediniospores in wheat field air,this paper focuses on the grading method of disease severity of wheat stripe rust,the segmentation and counting method of urediniospore microscopic images,and the remote real-time capture and microscopic image collection system of urediniospores in wheat field air.The monitoring and identification methods were proposed and improved to establish foundation for the ultimate realization of the prediction of the wheat stripe rust based on the Internet of Things in a wide area.In general,the major work,results and contributions of this dissertation are as follows.（1）To quickly and accurately evaluate the disease level of wheat stripe rust,a novel grading method of disease severity of wheat stripe rust based on hyperspectral imaging technology was proposed.Firstly,hyperspectral images of 320 infected at different levels and 40 healthy wheat leaf samples were captured by a HyperSIS hyperspectral system covering the visible and near-infrared region（400¹000 nm）.Secondly,via the analysis of spectral reflectance of leaf and background regions,there were obvious differences in spectral reflectance at the 555 nm wavelength.Therefore,the image of the 555 nm wavelength was named the feature image,which was manipulated by threshold segmentation to obtain a mask image.The logical and operation was conducted by using the original hyperspectral image and mask image to remove the background information.Thirdly,the principal component analysis（PCA）method was used for the dimension reduction of hyperspectral images.The operation results showed that the second principal component image（PC2）can significantly identify the stripe rust spot area and healthy area.On this basis,stripe rust spots area was efficiently segmented by using an Otsu method.Finally,the degree of the disease severity of wheat stripe rust was graded according to the proportion of stripe rust spots area on a whole leaf.To verify the effectiveness of the proposed method,a total of270 leaf samples were collected for the performance evaluation.Experimental results showed that 265 samples could be accurately classified at different disease severity of wheat stripe rust and the overall classification accuracy was 98.15%.In conclusion,the experimental results indicate that the method using hyperspectral imaging technology proposed in this study is able to satisfy the precision demand of quantitative calculation and provide a foundation to evaluate the field disease level of wheat stripe rust and a new idea for resistance identification method of wheat stripe rust.（2）To solve the problem of the over-segmentation of the urediospores induced by several local minima after distance transform by watershed algorithm,an improved segmentation and counting method based on the result of distance transform is proposed.Firstly,the color urediospore target areas image was segmented by K-means clustering algorithm from the original image,and converted into a binary image.The morphological processing including region filling,small area removing and open operations were implemented to fill the spores,filter the noises and smooth the spore’s edges.Secondly,the local minimum was extracted from the gray-scale image of urediospores after distance transformation,and the local minimum was taken as the local minimum of the gradient image.Then,the watershed algorithm was used to segment the touching urediospores.The segmentation and counting experiments of 3857 urediospores were carried out.The accuracy rate reached 92.6%.The experimental results showed that the proposed method got better segmentation and counting results,which solved the problem of over-segmentation of urediospores and achieved efficient and accurate counting of urediospores.（3）To improve segmentation and counting accuracy of touching objects in spores counting system,a segmentation and counting algorithm for urediospores based on concavity and contour segments merging was proposed.Firstly,shape factor that was defined using area with perimeter of a region and area were used to be parameters to justify whether the region was one touching region or not.Secondly,if the region was identified as a single spore,the ellipse number was directly counted by the least-squares ellipse fitting method.For the touching spores,the contour of touching spores was extracted and divided into contour segments（CS）with the concave spots on it.The next step was to merge the contour segments belonging to the same uredispore.The distance measurement（DM）and deviation error measurement（DEM）were proposed to test whether the contour segments pertain to the same spores or not.If they both were satisfied the threshold values of distance measurementω_DM（40）and deviation error measurementσ_DEM（95）,which were merged as a new segment.With these newly merged contour segments,the ellipses are fitted as the representative ellipses for touching spores.Finally,the all ellipse number were added up as the number of urediospores.To verify the effectiveness of the proposed algorithm,the counting test was performed on 120 images of urediospores.The experimental results compared to that of the improved watershed segmentation algorithm showed that:The lowest counting accuracy of the proposed method was 92.7%,the highest counting accuracy was 100%.The total average counting accuracy was 98.6%,which was increased by 6.0%compared to the improved watershed segmentation algorithm（92.6%）.In conclusion,experimental results show that the proposed method is efficient and accurate for the automatic detection and counting for urediospores,which provide technical support for the development of on-line urediospores monitoring equipment.（4）Collecting airborne urediospores in wheat fields using spore trap devices,has currently become an important approach for devising strategies early and effectively controlling wheat stripe rust.However,the existing spore trap devices have some shortcomings,such as low efficiency,time-consuming,and requiring manual replacement of slides or plastic tapes.To this end,a high-magnification and high-resolution micro-image remote acquisition system of uredinispores was designed and manufactured based on ARK-1123C embedded industrial computer and microscope digital camera.The hardware and software architectures of the micro-image remote acquisition system were designed,which realized a series of functions including automatic supply of slide,coating a thin film of petroleum jelly,aerial urediniospore capture,urediniospore micro-image acquisition,and slide recycling.Besides,the parameters of urediniospore capture and micro-image acquisition could be set remotely according to user’s requirement,and the collected images could be transmitted to the remote server by 4G wireless network.In order to verify the performance of the system,the system was deployed and tested in a wheat field for 40 days.Experimental results showed that the system worked stably for a long time,and could capture the micro-images of uredinispores,with 400 times magnification and 4096-pixels×3288-pixels resolution.The experiments validate the remote acquisition system proposed can automatically collect and remotely transmit the microscopic image of uredinispores in real time,which provide important technical support for the automatic counting of the airborne uredinispores and the prediction of the wheat stripe rust in the field.It can also provide a reference for monitoring other airborne fungal spores.（5）A software for segmentation and counting of urediospores of wheat stripe rust was developed.In order to realize the automatic counting of urediospores,an automatic counting software for urediospores based on image processing was developed using MATLAB GUIDE platform in combination with Local C Compiler.The software was independent of the MATLAB environment and can be run on a computer without MATLAB software.Using this software,automatic counting of urediospores in a microscopic image can be implemented via image processing technologies including image scaling,K-means clustering segmentation,morphological processing,the improved watershed segmentation and counting algorithm,the segmentation and counting algorithm for urediospores based on concavity and contour segments merging,etc.Structure design of the automatic counting system,the key algorithms used in the system and realization of the main functions of the system were described in detail.By using the automatic counting software for the urediospores counting test,the average counting accuracy was more than 95%.The results showed that the software could count the urediospores quickly and accurately,which provided a convenient and accurate tool for the on-line monitoring system of urediospores of wheat stripe rust in the field.