Research On Early Detection Mechanism Of Wheat Kernels Infection And Reconstruction Of Wheat Kernels Images Based On Micro-CT

The losses caused by pest infection in China's farmers' stored kernels are about 8~10%,and the direct economic losses exceed 2 billion yuan every year.The grainworm is the most harmful in the larval stage.If the pests inside the kernels are found in the early stage and detected accurately and automatically,it can provide a scientific basis for early reasonable control.However,various current detection techniques are not able to automatically detect wheat grains containing eggs and young larvae.Micro-CT equipment can obtain high-resolution projection data of wheat kernels,and analyze the influence of infestation on the internal microstructure of wheat kernels to realize the automatic detection of pest infection inside the wheat kernels.This paper focuses on the early detection mechanism of wheat kernels insect damage based on Micro-CT,the Micro-CT image reconstruction of wheat kernels and the correction of ring artifacts.The mechanism of early detection of wheat kernels pest infection based on micro-CT was discussed.In the immature stage of Sitophilus oryzae,the feeding and development in the kernels,the protein and starch in the grain are consumed,and the content of the main components in the grain changes,which leads to the change of internal structure and density after the wheat kernels are infested.It can be seen from the scatter plot that the density and starch of wheat kernels increase with the development of Sitophilus oryzae,and the protein and starch show a downward trend.Different densitie substances have different attenuation rates for X-rays,so the characterization strength of healthy kernels and the infested kernels on micro-CT images is different.The results show a very significant linear correlation between wheat kernels density and starch content,and the correlation coefficient is 0.971,which had a significant negative linear correlation with protein and water content.There is a very significant linear correlation between the micro-computed tomography images gray value and density of wheat kernels,and the correlation coefficient is 0.988.Therefore,the CT imaging system can detect changes in the internal structure information of the kernels after infection by the pest.A three-dimensional cone beam reconstruction model of wheat kernels was constructed.The hairy end and embryo end of wheat kernels images reconstructed by FDK algorithm produce serious artifacts,which will blur the Sitophilus oryzae insect path and affect the detection effect of kernels pest infection.Two kinds of kernels images reconstruction algorithms,Weight-FDK and z-FDK,based on back-projection weight function are studied,and the three aspects of the adaptability of algorithms,the correction effect of artifacts and the quality of reconstructed images are compared.The Weight-FDK algorithm needs to optimize two parameters,which limits the applicability of the algorithm.The Weight-FDK and z-FDK algorithms improve the axial artifacts generated by the FDK algorithm to a certain extent.From the profile curve,z-FDK algorithm improved gray level more at both ends of wheat kernels more,and makes the middle area of the kernels and the two ends more consistent.The average gradient of z-FDK algorithm is 0.1488 and 0.4004 higher than that of Weight-FDK and FDK respectively,and the contrast noise ratio is 0.8683 and 2.4866 higher than that of Weight-FDK and FDK respectively.It can be seen that the z-FDK algorithm significantly reduces the artifacts of kernels images and improves the images quality.Aiming at the circular artifacts in wheat kernels reconstruction images,two correction methods,polar coordinate based ring correction and uniformity test ring correction,are studied.The results show that polar coordinate based on ring correction method eliminates part of ring artifacts in wheat kernels images,and the artifacts near the center of the images cannot be removed.The uniformity test ring correction method eliminates all ring artifacts in wheat kernels images.The grain area and the insect area in the images are more clear,which can more clearly characterize the pest infection information inside the kernels,which lays a good foundation for the study of the development of the kernels pest and the visual analysis of the infection status.