Research On Wheat Grain Image Reconstruction And 3D Visualization Based On Micro-CT

Nowadays,the household food stocks accounting for about 50% of the total grain output in China,the corrosion loss of wheat have been 2.61%-3.57% during the storage period of farmers,the annual wheat damage caused by pest damage have been 1.5-6 million tons,the direct economic loss has been more than 2 billion yuan.Over the past decade or so,domestic and foreign scholars had used a variety of new technologies to carry out a more in-depth study on the detection of grain pests.However,the detection of eggs and young larvae in grains was not satisfactory.The difference between the internal changes in intact grain and early invaded grain is very small.Micro-CT imaging system can obtain the micro-level resolution of grain stereoscopic information,achieving early automatic detection of internal grain pests by analyzing the grain microstructure changes.Therefore,studying on the accurate reconstruction and 3D visualization of grain is of great significance to the extraction of three-dimensional grain characteristics and identification.The main contents of this paper were the pretreatment of grain two-dimensional projection data,the reconstruction of wheat grain slice and the three-dimensional visualization of wheat grain.Setting the radiation source voltage of 45 k Vp,the current of 88 mA,the resolution of 14.8 mm and other scanning parameters to collect the two-dimensional projection of wheat grain image,the size was1004?500;In order to segment the scanning area which was beneficial to the reconstruction of wheat grain fault,the effective scanning region was automatically extracted by threshold segmentation method,the threshold was 220 and the size was 580?500;In order to remove the noise in the wheat grain projection images,the effective scanning data were denoised by gaussian filtering,mean filtering and median filtering.Finally,four wheat projection data,the gaussian filtered data,the mean filtered data,the median filtered data and the raw scanning data,were obtained for wheat slice reconstruction.The projection data of the three-dimensional head model was simulated and reconstructed based on the three-dimensional Shepp-Logan head model and the principle of parallel beam and fan beam scanning and reconstruction.Four kinds of grain projection data were reconstructed by cone-beam FDK algorithm,the size of reconstruction was 512?512.The results of four reconstructions were evaluated by noise intensity and contrast.The results showed that the reconstructed results of mean filtered-projection data were relatively good.The grain projection data of the mean filter were reconstructed in the FDK algorithm using R-L,S-L,Hamming and Blackman filters,respectively.The results showed that the quality of wheat grain slices reconstructed by Blackman filter is better in FDK algorithm.The data of wheat grain slices reconstructed by FDK were pretreated by threshold binarization,corrosion,fusion,smoothing and edge enhancement,and obtaining the suitable wheat grain slice data for three-dimensional reconstruction.The MC algorithm was used to reconstruct the grain data of the wheat grain after the pretreatment,and the three-dimensional drawing of the grain was realized.The threshold of the algorithm was 30,and the transparency was 0.2.The results showed that the surface rendering of wheat grain could observe the surface characteristics of wheat grain in all directions,and it could clearly see the change of wheat grain,which laid a good foundation for the extraction and identification of wheat grain.