eokeeffe / ssba Goto Github PK
View Code? Open in Web Editor NEWSSBA-V3.0 library
License: GNU Lesser General Public License v3.0
SSBA-V3.0 library
License: GNU Lesser General Public License v3.0
Description This is an implementation of a sparse Levenberg-Marquardt optimization procedure and several bundle adjustment modules based on it. There are three versions of bundle adjustment: 1) Pure metric adjustment. Camera poses have 6 dof and 3D points have 3 dof. 2) Common, but adjustable intrinsic and distortion parameters. This is useful, if the set of images are taken with the same camera under constant zoom settings. 3) Variable intrinsics and distortion parameters for each view. This addresses the "community photo collection" setting, where each image is captured with a different camera and/or with varying zoom setting. There are two demo applications in the Apps directory, bundle_common and bundle_varying, which correspond to item 2) and 3) above. The input data file for both applications is a text file with the following numerical values: First, the number of 3D points, views and 2D measurements: <M> <N> <K> Then, the values of the intrinsic matrix [ fx skew cx ] K = [ 0 fy cy ] [ 0 0 1 ], and the distortion parameters according to the convention of the Bouget toolbox: <fx> <skew> <cx> <fy> <cy> <k1> <k2> <p1> <p2> For the bundle_varying application this is given <N> times, one for each camera/view. Then the <M> 3D point positions are given: <point-id> <X> <Y> <Z> Note: the point-ids need not to be exactly from 0 to M-1, any (unique) ids will do. The camera poses are given subsequently: <view-id> <12 entries of the RT matrix> There is a lot of confusion how to specify the orientation of cameras. We use projection matrix notation, i.e. P = K [R|T], and a 3D point X in world coordinates is transformed into the camera coordinate system by XX=R*X+T. Finally, the <K> 2d image measurements (given in pixels) are provided: <view-id> <point-id> <x> <y> 1 See the example in the Dataset folder. Performance This software is able to perform successful loop closing for a video sequence containing 1745 views, 37920 3D points and 627228 image measurements in about 16min on a 2.2 GHz Core 2. The footprint in memory was <700MB. Requirements Solving the augmented normal equation in the LM optimizer is done with LDL, a Cholsky like decomposition method for sparse matrices (see http://www.cise.ufl.edu/research/sparse/ldl). The appropriate column reordering is done with COLAMD (see http://www.cise.ufl.edu/research/sparse/colamd). Both packages are licensed under the GNU LGPL. This software was developed under Linux, but should compile equally well on other operating systems. -Christopher Zach ([email protected]) News for SSBA 2.0 * Added a sparse LM implementation (struct ExtSparseLevenbergOptimizer) handling several least-squares terms in the cost function. This is useful when several types of measurements (e.g. image feature locations and GPS positions) are available. See Apps/bundle_ext_LM.cpp for a simple demo. * Changed the default update rule for the damping parameter lambda to a simpler one (multiply and divide lambda by 10, depending on the cost function improvement). This seems to work better that the more complicated rule used before. * Fixed a trivial, but important bug in cost evaluation after the parameter update. /* Copyright (c) 2011 Christopher Zach, Computer Vision and Geometry Group, ETH Zurich This file is part of SSBA-2.0 (Simple Sparse Bundle Adjustment). SSBA is free software: you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version. SSBA is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details. You should have received a copy of the GNU Lesser General Public License along with SSBA. If not, see <http://www.gnu.org/licenses/>. */
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.