3D Reconstruction through Measure Based Image Selection

3D reconstruction is one of the research focus of computer vision and widely applied in various fields. The main steps of 3D reconstruction include image acquisition, image selection, feature point extraction and matching, calculation of camera parameters and 3D coordinates of the scene, and production of dense 3D scene models. Of all the steps, the image selection step is necessary and important. In this paper, a new 3D reconstruction algorithm consisting of four main steps is proposed and its main contribution is that the image selection step uses a new effective method. The new image selection method first uses structure-from-motion (SFM) algorithms to calculate the position and attitude of each camera, and then calculates the contributed value to 3D reconstruction of each image, finally selects images according to the contributed value of each image and their effects on contributed values of other images. Experimental results show that the 3D reconstruction algorithm proposed in this paper can reconstruct target scenes very well.