Multi-view structure from motion (SfM) estimates the position and orientation of pictures in a common 3D coordinate frame. When views are treated incrementally, this external calibration can be subject to drift, contrary to global methods that distribute residual errors evenly. We propose a new global calibration approach based on the fusion of relative motions between image pairs. We improve an existing method for robustly computing global rotations. We present an efficient a contrario trifocal tensor estimation method, from which stable and precise translation directions can be extracted. We also define an efficient translation registration method that recovers accurate camera positions. These components are combined into an original SfM pipeline. Our experiments show that, on most datasets, it outperforms in accuracy other existing incremental and global pipelines. It also achieves strikingly good running times: it is about 20 times faster than the other global method we could compare to, and as fast as the best incremental method. More importantly, it features better scalability properties.

Abstract. The OpenMVG C++ library provides a vast collection of multipleview geometry tools and algorithms to spread the usage of computer vision and structure-from-motion techniques. Close to the state-of-the-art in its domain, it provides an easy access to common tools used in 3D reconstruction from images. Following the credo "Keep it simple, keep it maintainable" the library is designed as a modular collection of algorithms, libraries and binaries that can be used independently or as bricks to build larger systems. Thanks to its strict test driven development, the library is packaged with unit-test code samples that make the library easy to learn, modify and use. Since its first release in 2013 under the MPL2 license, OpenMVG has gathered an active community of users and contributors from many fields, spanning hobbyists, students, computer vision experts, and industry members.

The RANSAC [2] algorithm (RANdom SAmple Consensus) is a robust method to estimate parameters of a model fitting the data, in presence of outliers among the data. Its random nature is due only to complexity considerations. It iteratively extracts a random sample out of all data, of minimal size sufficient to estimate the parameters. At each such trial, the number of inliers (data that fits the model within an acceptable error threshold) is counted. In the end, the set of parameters maximizing the number of inliers is accepted. The variant proposed by Moisan and Stival [7] consists in introducing an a contrario [1] criterion to avoid the hard thresholds for inlier/outlier discrimination. It has three consequences: 1. The threshold for inlier/outlier discrimination is adaptive, it does not need to be fixed. 2. It gives a decision on the adequacy of the final model: it does not provide a wrong set of parameters if it does not have enough confidence. 3. The procedure to draw a new sample can be amended as soon as one set of parameters is deemed meaningful: the new sample can be drawn among the inliers of this model. In this particular instantiation, we apply it to the estimation of the homography registering two images of the same scene. The homography is an 8-parameter model arising in two situations when using a pinhole camera: the scene is planar (a painting, a facade, etc.) or the viewpoint location is fixed (pure rotation around the optical center). When the homography is found, it is used to stitch the images in the coordinate frame of the second image and build a panorama. The point correspondences between images are computed by the SIFT [5] algorithm.

No abstract