KalNet Message Board

This software (CMVS) takes the output of a structure-from-motion (SfM) software as input, then decomposes the input images into a set of image clusters of managable size. An MVS software can be used to process each cluster independently and in parallel, where the union of reconstructions from all the clusters should not miss any details that can be otherwise obtained from the whole image set. CMVS should be used in conjunction with an SfM software Bundler and an MVS software PMVS2 (PMVS version 2). — Read more
http://opensourcephotogrammetry.blogspot.com/

This project intend to create a python photogrammetry toolbox.
It provides an easy interface to run Bundler + Dense point cloud computation via PMVS2 and CMVS (as WIP).

The main drawbacks of Bundler is that people have to install cygwin to use it and windows.
The advantage of Python is that the scripting langage is multiplatform, so the same code will be ok to run Bundler and the other tools on windows, linux and mac ! No more cygwin... installation.

This project make the Sift detector from VLFeat library works with Bundler. So the toolchain is OpenSource from A to Z. Other modules could be plug into because of the modularity approach that have been choosen.

In a near future we could think in integrate a SIFTGpu matcher into the toolchain.

VisualSFM is a GUI application for 3D reconstruction using structure from motion (SFM). The reconstruction system integrates and improves upon several of my previous projects: A course project to implement a PhotoTourism-like incremental SFM system, SIFT on GPU(SiftGPU), and the recent Multicore Bundle Adjustment. VisualSFM is able to run very fast by exploiting multicore parallelism in feature detection, feature matching, and bundle adjustment.

In addition, VisualSFM provides the interfaces to run Yasutaka Furukawa's PMVS/CMVS tool and to prepare data for Michal Jancosek's CMP-MVS. This software also comes with two research projects: Schematic Surface Reconstruction and Single View Repetition Analysis.

Photofly is a open source photogrammetry software which works well for modeling objects. When it comes to buildings, its goal does not seem to be the final creation of models, as Imagemodeler was. Instead it seems more intended for recreating an existing space for reference that you will add Autocad designs to. People briefly used the original version (Photofly version 1) with mixed success.
Some testing result of photogrammetry software Photofly is given below. It's an interesting piece of software with pretty good results.

Structure from Motion Toolkit is a open-source photogrammetry software with complete photogrammetry solution. SFM_georef is a gui-based Matlab tool for scaling and geo-referencing structure-from-motion (SfM) point clouds to real-world coordinates, using observations made directly in the SfM image set (rather than identifying and matching features from the point cloud).

EnsoMOSAIC photogrammetric software is a great application for its price. New users interested in the world of 3D aerial photogrammetry should find a good place to start with this application. Editing and image processing in EnsoMOSAIC is easy, fun, and headache-free. This photogrammetry software has always been famous for its rendering engine one of the most multiprocessor aware, multi-threaded implementations of any application, on any platform.

insight3d lets you create 3D models from photographs. You give it a series of photos of a real scene (e.g., of a building), it automatically matches them and then calculates positions in space from which each photo has been taken (plus camera’s optical parameters) along with a 3D pointcloud of the scene. You can then use insight3d’s modeling tools to create textured polygonal model.

Basically, it hopes to be "Photosynth + Canoma". Or "boujou + Imagemodeler". But it's also free and opensource (GNU AGPL 3). Check out the screenshots below or download it already.

The Open Source Geospatial Foundation...
OSGeo was created to support the collaborative development of open source geospatial software, and promote its widespread use. Join us by signing up to our mailing lists or check out the Getting Started page to become more involved.

Get started here: gettingStarted

This toolbox is a compilation of popular Structure From Motion (SFM) algorithms. As I used more and more SFM over the years, I could never find anything that goes beyond simple SFM (please, prove me wrong and send me links !!). I compiled all my code into a toolbox so that others do not have to reimplement everything. I apologize in advance if you do not find what you need (like algorithms using the L-infinity beauty) but I only implemented what I needed and optimized what I needed most:

data generation (random rigid/non-rigid scene, popular non-rigid examples, different camera models, random or smooth random camera positions (with circle splines for Sequin), noise addition ...)
2D geometry (homography computations ...)
3D geometry (quaternion, globally optimal pose estimation, absolute orientation, triangulation ...)
rigid 3D reconstruction: orthographic (Tomasi-Kanade with or without orthonormality constraints) or projective (eight point, Sturm - Triggs 96, Oliensis - Hartley 07, globally optimal affine/metric upgrades from Chandraker 09 (with bug fixes to the original papers)), bundle adjustment ...
non-rigid 3D reconstruction: interface to the available Torresani 08, optimized Xiao - Kanade 04 (with bug fixes to the original papers), CSFM from Rabaud 09
visualization for rigid/non-rigid (sequences , multiple-views at once, comparisons ...)
This toolbox is in development and is maintained by Vincent Rabaud (http://vision.ucsd.edu/~vrabaud/). Thanks to everyone who has given me feedback and suggestions you've helped make this toolbox more useful and easier to use.

If you use this toolbox in a paper, I would be grateful if you could cite it using the following BibTex entry:

UAV data accuracy checks. Wait, what?
Jun 8
Written By Josh Tremba
Scrolling through endless drone mapping forums, message boards, youtube channels, you name it (deep down a rabbit hole) it began to dawn on me that a reliance on internal accuracy checks was not only an adopted technique, but a widely accepted and even recommended modus operandi. To a data geek and accuracy wolf that shook me a bit. I could not imagine relying on internal (i.e. software derived bundle adjustment checks) error estimates to certify a map for any type of actual planning. Even landscaping plans. How will those palm trees look in 10 years when your map had a scale factor of .956548? Probably fine, but I enjoy perfectly spaced palm trees.

The Federal Geodetic Data Commission (FGDC) is the head honcho in the USA right now with regards to accuracy standards. They created the “Geospatial Positioning Accuracy Standards" parts 1-5 (part 3 really gets into the fun stuff, and in part 5 I think the terminator travels to the future from the past for once). The gist of the story is, you must do independent checks on your mapping with methods of measurement that have higher accuracy than which you used to make a map. That means if you mapped with GCPs and a drone, you must take measurements independent of your control to create a proper map check. And yes, even if you measured your GCPs with RTK, doing a map check with RTK is completely acceptable because the method of measurement for your final map is not RTK but rather a photogrammetric bundle adjustment (done in your flavor of Pix4D/Meatashape) which inherently has more error than the control it uses.

So give this a whirl; take 20 or so random RTK shots on your next UAV mapping project, and calculate a plus/minus for each one relative to your final mapping surface. Square each of those plus/minuses and then add them all up. Now take the square root of that sum (standard deviation if you haven’t noticed), and multiply it by 1.98 to get your 95% confidence error. What that means is statistically your map will fall plus/minus that number below or above the truth 95% of the time. You have an independent accuracy number, yay! Now see what Pix4D told you…