In order to solve the problem of illumination which leads to poor segmentation effect of rapeseed image in the field, the method of combining EXG algorithm and Lab space threshold segmentation algorithm is used to realize the location of all rapeseed regions by EXG color segmentation. However, due to illumination problem, the color similarity between the flower stalk and rapeseed leads to false segmentation. The rapeseed image in RGB color space was transformed into Lab color space, and the b-component was threshold segmented, but the green part of the blooming rapeseed flower and bud was undersegmented. Therefore, combined with EXG algorithm and Lab color space b-component threshold algorithm, the accurate segmentation of rapeseed flowers in field rapeseed crops is realized. The mean value and variance of segmentation error are only 8.33%, 0.22% and the mean value and variance of classification error are only 0.67%, 0.007%, which are lower than those of the other two algorithms. The results of rapeseed image processing in different regions show that the combination of EXG algorithm and b-component threshold segmentation in Lab color space can extract rapeseed flowers accurately, which effectively avoids the problem of false segmentation of flower stalks and rapeseed flowers caused by illumination, and undersegmentation of the green part of rapeseed flowers and buds.