Study On The Recognition And Larval Development Of Sitophilus Oryzae In Wheat Kernel Based On Computer Vision

China is a large producer of grain,the planting area of wheat reaches about 22% of the total grain planting area,which accounts for a large proportion in grain production.In China,the direct economic losses caused by insect amount to more than 2 billion yuan each year,which poses a great hidden danger to grain security.Therefore,the control of insects is particularly urgent and important.In recent years,many experts and scholars have continuously researched deeply and achieved certain results in the detection of stored grain pests,but the detection of early insects in wheat kernels has not achieved the desired results.In this paper,the Micro-CT imaging equipment was used to achieve the early detection of stored grain pests in wheat kernels,and the method based on machine vision for research on the abnormal development of Sitophilus Oryzae in wheat kernels were proposed.The 3,9,17,22,and 28 days after the adult Sitophilus Oryzae and wheat kernels mixed are the early larval stages,young larval stage,high larval stage,pupal larval stage and adult larval stage.The infected wheat kernels at different larval stage were scanned by Micro-CT to obtain the RSQ data volume.The z-FDK algorithm was used to reconstruct the images of infected wheat kernels in different larval stage after analyzing the projection data.The images obtained were filtered,and the otsu method were used to segmentation the infected wheat kernels images to obtain binary images of Sitophilus oryzae.The connected area method was used to label the image to obtain different connected areas in the image,and the small areas caused by dust and noise were removed.The image processing method was used to remove the circular sampling tube and wheat kernels in the image,the regions of interest in the image were extracted,and the volume data of infected wheat kernels and Sitophilus oryzae in different stages were constructed.According to the volume data constructed,the twenty-six features were extracted to construct the original feature space,including eight two-dimensional features,four threedimensional features,seven moment-invariant features,and seven salient texture features based on the gray-level co-occurrence matrix.Taking six morphological characteristics such as volume,surface area,height,and maximum cross-sectional area as examples to analyzing the development of Sitophilus Oryzae in wheat kernels.The formula of the volume,surface area and other features changing with the development time were constructed.The simulated annealing algorithm was used to optimize the constructed 26-dimensional original feature space.When the fitness function value reaches a maximum of 90.21%,there are 10 features were selected.The artificial bee colony algorithm was used to optimize the penalty factor and radial basis kernel function parameters of the support vector machine,the extreme learning machine was also used for larval stage recognition.The results show that the development of Sitophilus Oryzae in wheat kernel is consistent with the actual situation,and the recognition rate of Sitophilus Oryzae reaches 97% when c=96.44,g=0.01.It shows that it is feasible to use machine vision to distinguish the different larval stages of Sitophilus Oryzae.