Pierre Moulon

Global Fusion of Relative Motions for Robust, Accurate and Scalable Structure from Motion

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.

Automatic Homographic Registration of a Pair of Images, with A Contrario Elimination of Outliers

The RANSAC [2] algorithm (RANdom SAmple Consensus) is a robust method to estimate parameters of a model tting 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 sucient to estimate the parameters. At each such trial, the number

Adaptive structure from motion with a contrario model estimation

Structure from Motion (SfM) algorithms take as input multi-view stereo images (along with internal calibration information) and yield a 3D point cloud and camera orientations/poses in a common 3D coordinate system. In the case of an incremental SfM pipeline, the process requires repeated model estimations based on detected feature points: homography, fundamental and essential matrices, as well as camera poses. These estimations have a crucial impact on the quality of 3D reconstruction. We propose to improve these estimations using the a contrario methodology. While SfM pipelines usually have globally-fixed thresholds for model estimation, the a contrario principle adapts thresholds to the input data and for each model estimation. Our experiments show that adaptive thresholds reach a significantly better precision. Additionally, the user is free from having to guess thresholds or to optimistically rely on default values. There are also cases where a globally-fixed threshold policy, whatever the threshold value is, cannot provide the best accuracy, contrary to an adaptive threshold policy.

OpenMVG: Open Multiple View Geometry

The OpenMVG C++ library provides a vast collection of multiple-view 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.

Fundamental Matrix of a Stereo Pair, with A Contrario Elimination of Outliers

In a stereo image pair, the fundamental matrix encodes the rigidity constraint of the scene. It combines the internal parameters of both cameras (which can be the same) and their relative position and orientation. It associates to image points in one view the so-called epipolar line in the other view, which is the locus of projection of the same 3D point, whose particular position on the straight line is determined by its depth. Reducing the correspondence search to a 1D line instead of the 2D image is a large benefit enabling the computation of the dense 3D scene. The estimation of the matrix depends on at least seven pairs of corresponding points in the images. The algorithm discarding outliers presented here is a variant of the classical RANSAC (RANdom SAmple Consensus) based on a contrario methodology and proposed first by Moisan and Stival in 2004 under the name ORSA. The distinguishing feature of this algorithm compared to other RANSAC variants is that the measure of validity of a set of point pairs is not its sheer number, but a combination of this number and the geometric precision of the points.

On the determination of 3D trajectory of moving targets by stereovision

This paper describes a vision system to obtain the 3D trajectory of a moving target by stereovision. For our study, we chose to reconstruct the trajectory of a clay plate that is used in ball-trap. The plates path can reach 60 meters in length. The trajectory is reconstructed from videos of the moving target and plate positions between the left and right images at different times. To find the 3D coordinates of the plate, we use a triangulation algorithm on couple of points that were found prior. To obtain the whole trajectory of the object, we use a spline interpolation. Our stereovision system is very simple, it uses two classical webcams with a resolution of 640*480 pixels at 30 frames per second. We present in this paper the architecture of the developed system and show its efficiency through experimental results. We conclude with some perspectives for improvement in future work.