Abstract: 3D structure of maize leaves is an important indicator for evaluating biological characteristics. In order to generate 3D structure of maize leaves rapidly, simply, and accurately, the authors proposed a 3D reconstruction system mainly consisting of two high resolution cameras. The chessboard was chosen to be the plane calibrating template mark, and structured light was applied in stereo matching process to calculate 3D point position. After interpolating leaf 3D points with Cardinal spline and triangulating them into surfaces, a part of leaf 3D structure could be acquired, and then translating and rotating the different parts of maize leaf to form the final complete structure. Experimental results of 3D reconstruction of maize leaves show that this method has the advantage not only of high precision but also of non-contact, non-destructive, and automatic operating process.