The three-dimensional (3D) reconstruction technology based on machine vision is researched. We use the pictures shot by ordinary digital cameras to achieve 3D reconstruction through camera calibration, feature-point detection and matching, and the calculation of fundamental matrix and essential matrix. We complete the camera calibration by Camera Calibration ToolBox for Matlab based on Zhang Zhengyou′s method. The scale invariant feature transform (SIFT) method is used for image feature-points detection and matching, and the random sample consensus (RANSAC) algorithm is adopted to calculate the fundamental matrix and the essential matrix. Then we use the camera internal parameters got in advance and the essential matrix obtained from the fundamental matrix to reconstruct the feature points of the objects and achieve the 3D effect by pasting pictures of texture. The experimental results show that it can complete the reconstruction and can well reflect the 3D feature of the objects.