Three-dimensional shape descriptors of corn ears are important traits in corn breeding, genetic and genomics research, however it is difficult to accurately and consistently measure 3D features of corn ears by hand or traditional tools. This study presents a 3D modeling method based on machine vision to reconstruct the 3D model of corn ears for quantitative feature computation and analysis. Firstly, a simple machine vision system is designed to capture images of corn ears from different angles of view. The corn ears in these images are then registered in the uniform coordinate system using a rapid process pipeline which consists of image processing, object detection, distortion correction and registration in pixel level etc. After the registration, the point sets in edge contours and center skeletons of corn ears are used to reconstruct the surface model based on resample and interpolation techniques. The experimental results demonstrate that the presented method can not only build realistic 3D models of corn ears for visualization, also be used to accurately compute geometric characteristics.