In order to solve the problem of inaccurate extraction of internal components in maize grain transmission image due to low contrast, a multichannel overlapping region segmentation method based on nonlinear transformation of color channel was proposed. The horny endosperm pixel information was obtained from the binary images of gray scale and R channel of maize; and the “farinaceous endosperm+embryo” image was transformed into Lab color space, the farinaceous endosperm image under b channel was extracted; and then the embryo pixel information was obtained from the binary images of R channel and b channel, so as to segment the components of maize grain accurately. Firstly, the transmission images of different maize varieties were collected, and the single grain was extracted. Secondly, the single maize grain images were segmented by the multichannel overlapping region method based on the nonlinear transformation of color channel, and the images of horny endosperm, farinaceous endosperm and embryo of maize grain were obtained. Finally, the recall ratio and the precision ratio were used to evaluate the segmentation effect of multichannel overlapping region method and traditional image segmentation methods on different varieties of maize seeds. The experiments showed that the recall ratio, precision ratio and comprehensive evaluation index of the multichannel overlapping region segmentation method were all over 98%. The segmentation effect was better than that of the traditional segmentation methods. It can realize the accurate segmentation of the transmission images of different varieties maize grain with transparent horny, and provide a scientific basis for the study of the threedimensional distribution of internal components and the mechanism of dry crack formation in maize grains.