To guide the formation mechanism of dry cracks inside maize kernels, a method is proposed to accurately measure the density of maize kernel components. Firstly, the maize kernels are grinded in layers by a layering grinding device. Then, after each grinding, the images of the grinded surface and side of the maize kernels are collected, and the part of the maize kernels removed by each grinding is regarded as the grinding piece, and the mass of the maize kernel grinding piece is measured; Secondly, the images of the down side of grinding piece are segmented by a K-Means clustering mean algorithm into 3 parts——keratinous endosperm, the farinaceous endosperm and the embryonic part. Next, the actual areas of the three parts and the height of the grinding pieces are measured, and we obtain the volume of each components; Finally, the density of each maize kernels components is obtained by a linear neural network model, and a test is validated by the quality of the maize kernels grinding pieces. The test results show that the accuracy of the measurement method reaches more than 96%, which can accurately measure the density of the internal components of different varieties of maize kernels, and provide the basic theory for the formation mechanism of dry cracks inside maize kernels.