Abstract: Abstract: The accurate detection about early insect pest inside grains can determine the infection of the grain in advance, which has great significant for developing reasonable control measures. This paper proposed a method for non-destructive detection of early infestation of wheat kernels based on Micro-CT system. The 3rd, 9th, 17th, 22nd and 28th day after the mixture of Sitophilus oryzae Linne and the perfect grain, were named respectively the egg stage, the young larva stage, the elder larva stage, the pupa stage and the adult stage under the experimental conditions. A series of micro-CT projection data was randomly selected near the middle of the grain for gray scale statistics, and the oscillation amplitude and gray scale difference were used as the basis for judgment. And the optimal parameter combination of the Micro-CT grain detection system was determined to be 55 kV anode voltage and 0.1 mm aluminum filter. When the main component of the grain changed by 1-6 mm, as the thickness of the grain decreased, the average gray value of the gray cube at the equidistance from the reference surface of the grain end of the grain gradually increased. That is, when the thickness of the grain was small, the path through which the X-ray passed was short, and the average gray value of the obtained projected image was increased. The change of the length of the X-ray through the path of the wheat grain caused a change in the average gray value of the projected image of the grain, and the shorter the path through which the X-ray passed, the larger the average gray value of the projected image. The aspect ratio of the grain was 1.53-4.17. When 360° rotation imaging, the different attitudes during imaging had a great influence on the length of the X-ray passing through its path. In the two-dimensional projection imaging of wheat grain, as the angle between the long axis of the grain and the rotating table increased, the average path length of the X-ray through the grain decreased, and the rate of change of the path length decreased, therefore the average gray value of the projected image increased and improved the consistency of the grain projection imaging. As a result, it was determined that the optimum scanning angle was that the long axis of the grain is oriented at 90° to the rotating table. Statistics the projection data of the infected grains in tests, through the slice data reconstructed by FDK algorithm, it could be seen that, the cross-sectional appearance of the grain image changed from a single circle to an oval shape and then to a multiple combination of circle or oval shape, while the spacing between the rice and wormholes expanded gradually, and the wormholes erode from the epidermis to the grain groin and then expanded around epidermis. The three-dimensional visualization data drawn by Mimics platform showed that, during the development of the grain image, the appearance of the grain image changed from rod to ellipsoid and then each organ developed, when the volume of the grain image arrived to its maximum in the pupa stage and the wormhole gradually extended from the epidermis to the center of the grain and eventually extended through the whole grain. The results showed that the non-destructive testing using Micro-CT system could accurately express the changes of the internal microstructure of wheat grains, and provided evidences for the study of early infective grains.