Abstract: Color is an important appearance quality of wheat, and is a main feature for wheat variety recognition and quality detection based on machine vision. At present, the research on the time-varying characteristics of stored wheat color is not deep enough. Therefore, the research on the color variation in each channel of wheat kernel images during storage was carried out. Three batches of images of wheat kernels from the same batch were collected at an interval of three months, and the mean, distribution, clustering and classification performance of color values were investigated. The results showed that the color values of each channel basically were in normal distribution; with the increase of storage time, the average values of red, green, blue, and gray increased, the peak values of probability density of green, blue and gray values shifted to the right, and the peak values of probability density of red values maintained at the far right side and increased rapidly; the color of wheat grains images at three time points showed significant clustering characteristics and classifiability, which indicated that the characteristics of wheat changed obviously.