Activity for Dorit Borrmann
-
Dorit Borrmann committed [r2378]
initialize pairing mode
-
fixed autoOct for scanio with data_suffix other than .3d
-
Added option toggle off show display objects
-
cleanup CMakeLists.txt
-
Compute lenght of trajectory based on pose or frames files
-
Added default behaviour for -s and -e to scan_red (start at 0 and end at last available scan)
-
Hi, I'm sure, JoSch could answer most questions better than me, but I'll give it a try anyways: yes, indeed. For that kind of data you should give the voxel size in cm. Regarding the masks, there is one small thing to consider. When reading the data some filters may be apllied. You can specify a minimum range and a maximum range, for example. I think there is also a peopleremover-specific filter applied to reduce the number of misdetections. The peopleremover only considers points which are not filtered...
-
Some more avoiding unnecessary normal calculation for point pair search.
-
Avoid unnecessary normal computation for point pair search.
-
Please note though, that there are a few important exceptions, most importantly the xyz and the rxp (proprietary format for Riegl laser scanners) format. As there are so many default parameters in all the programs in 3DTK some assumptions had to be made for those. The default format (uos) is in a left handed coordinate system and for the default parameters scenarios with mobile robots and data in cm are assumed. If your data is in a right handed coordinate system, you might want to use the xyz format....
-
maybe no more carriage return despite the fact that I couldn't find any before
-
merge gone wrong
-
Bug fix for normal calculation, needed in scan2scan distance
-
Added tool to compute scan2scan distance
-
removed debug output
-
-r parameter now working in show.
-
Compute distance of points in global coordinates instead of local coordinates.
-
syntax error
-
changed default control files to install no ROS related packages
-
fixed FIXME from scan::calcReduced()
-
added keyboard functionality for all 6 DoF motion in show
-
Hi Anton, for these none basic data types, you need to add a validate function. You can check it out, for example, in src/slam6d/exportPoints.cc it is implemented for the IOtype. I hope that helps! Best, Dorit
-
Hi Anton, there was recently a change to Surf in Feature Based Registration. Have a look at the changes in: https://sourceforge.net/p/slam6d/code/2236/ Please check, if you are using a current version of 3DTK and verify if your OpenCV version is compatible with the #if #else construct in the code. Best, Dorit
-
Hi Anton, are you talking about scan_to_panorama or scan_red? If you are talking about scan_to_panorama: First, please update to the current revision (2223). I just committed a small bug-fix for the ZAXIS projection method that removes hard coded limits in the code. Second, "-m 180 -w 180" sets the minimum and maximum horizontal angles to 180 degrees, resulting in a FOV of 0 degrees. "-m 0 -w 360" should lead to better results. Best regards, Dorit
-
removed hard-coded values for zMin and zMax for z-Axis projection
-
noanimcolor also for advanced option to iterate through frames
-
removed trailing whitespaces
-
Inverted checkbox for noanimcolor
-
-Added checbox for --noanimcolor to show
-
Added customFilter to caliboard
-
Fixed check for old rivlib
-
use all points in getPtPairsParallel
-
Fixed error in scan2utm; added some more (preliminary) functions for getting poses from GPS devices.
-
Optional delimantor for strtoarray
-
Added option to use reflectance information as gray scale for coloring meshes
-
Enabled ability to reduce scans directly in exportPoints
-
Corrected scan transformations for mesh reconstruction with reduction.
-
Updated mesh reconstruction to use aknn from slam6d/normals
-
fixed bug in recoverPointCloud used for range reduction
-
Adjusted some parameter descriptions in exportPoints
-
Adjusted scale factor for export points to work with both uos and xyz format.
-
Hi, to add to Josch's answer, as you noticed, the pose.dat file saves a pose in the viewer, including some additional parameters, such as the fog. Please refer to the savePose-function for more details in "src/show_show_animate.cc" The .pose files, however, store the initial pose of the scan, i.e., the transformation that transforms each individual scan into a global coordinate system. This is relevant, of you have more than one point cloud. Thus the .pose files transform the scans, while the 'pose.dat'...
-
-Add missing descriptions to boost::program_options in slam6d
-
Only compile curvefusion if WITH_ROS is set.
-
Fixing some of the issues with the boost::program_options
-
remove dummy source file from CMakelist.txt
-
scan_io_laz can now use an options file to enable options implemented in the lastools
-
Updated conversion between ECEF and local 3DTK coordinate system
-
add validate function for IOtype to scan2utm
-
restored linking XGetopt for programs that are not yet using boost::program_options
-
Hello Marek, if I interpret your data correctly, the scanner scans the profile, i.e. it is mounted in a way that it rotates around the axis of the movement of your UAV. This is a challenging setup for the global optimization. Just imagine you are flying over a completely planar surface, all your scans would look exactly the same. With the initial pose estimate these identically looking scans have a small (correct) offset, due to the limited field of view of the scanner. In the matching step, all...
-
The rotation matrices in the .frames files are always absolute. To get the final transformation for any scan you just need to take the last line from the corresponding .frames file.
-
Sure, but you need to modify the code for this. In file "src/show/show_animate.cc" remove or comment line 368: selectionfile << "# points from scan nr " << i << std::endl;
-
Hello Marek, if you use the option '--anim=1' and visualize the data with show afterwards, you can click on 'Animate' to the what the algorithm does in each iteration. This should help with identifying where the problem is. It could be with the prealignment (or your dataset in general) or with the parameters for the registration. Best regards, Dorit
-
draw coordinate systems correctly in advanced mode
-
bugfix for coordinate systems after animation
-
use ${EIGEN3_INCLUDE_DIR}
-
- add option to display the local coordinate axes of each scan
-
Hi, please adjust the following function in src/scanio/helper.cc to: bool ScanDataTransform_xyz::transform(double xyz[3], unsigned char rgb[3], float refl, float temp, float ampl, int type, float devi) { double tmp; tmp = xyz[2]; xyz[2] = 100.0xyz[0]; xyz[0] = -100.0xyz[1]; xyz[1] = 100.0tmp; return true; } That way your data will internally be treated as cm and the default parameters make more sense. Using NR=10 would be a good first try to reduce that data to one point per voxel with side length...
-
Hi, is your data in meters? Our standard data sets are all in cm and thus all the default parameters for 3DTK assume data to be in cm. This way all the points are so close together, that the algortihms simply fits a plane through the entire point cloud. If this is not the problem, try reducing your point cloud with the octree reduction to enforce equal density. Best regards, Dorit
-
use xfeatures only for opencv nonfree
-
remove namespace xfeatures
-
try to make non-free features work for OpenCV 3
-
basic tools for gps related things
-
moved check for color and reflectance to io_types
-
added coordinate systems to show display objects
-
Adjust installation instructions for Debian Jessie
-
Test commit
-
added coordinate systems to show display objects
-
Program to apply a transformation to all frames.
-
program to align scans with known correspondences
-
option WITH_SHAPE_DETECTION instead of WITH_PLANES
-
fix for ./bin/scan_red --reduction RANGE
-
Hi, the error hints to a missing X server on your Linux subsystem. As I don't have any way to try this, I did some research. From what I found, the most common options are: https://sourceforge.net/projects/xming/ or https://sourceforge.net/projects/vcxsrv/ or http://x.cygwin.com/devel/server/ After installing one of them, you need to setup the DISPLAY environment: export DISPLAY=:0 I am not sure if you need any other packages, like x11-common Best regards, Dorit On 04/29/2017 09:54 PM, Cameron wrote:...
-
Hi, the error hints to a missing X server on your Linux subsystem. As I don't have any way to try this, I did some research. From what I found, the most common options are: https://sourceforge.net/projects/xming/ or https://sourceforge.net/projects/vcxsrv/ or http://x.cygwin.com/devel/server/ After installing one of them, you need to setup the DISPLAY environment: export DISPLAY=:0 I am not sure if you need any other packages, like x11-common Best regards, Dorit
-
Add ANN requirement for MAC
-
Fix for compiling the compact octree
-
convert ply files from left to right handed (y = -y), to show files from vsfm nicely
-
export color for ply files
-
added BoxDisplay
-
Bug fix in Hough
-
Dear John, if I understand correctly, what you want to do, this is the general idea...
-
H
-
exportPoint: changed -h to -H
-
-exportPoint:
-
Fixed dependency of scan_io_b3d to WITH_TOOLS
-
reverted to reading frames from m_path/scan*.fr...
-
Restored original Makefile
-
HI, deleting all the points except the ones from the walls is essentially already...
-
Hi, planes is actually calculating planes only based on a single scan. In your case...
-
Hi, what is your command line call for getting these results? What are the dimensions...
-
Hi, please make sure, the parameters in bin/hough.cfg are compatible with your dataset....
-
Hello John, for most applications the chosen method has hardly any influence on the...
-
Hallo John, for most applications the chosen method has hardly any influence on the...
-
Hello Jhon, the methods are described in the paper cviu2010.pdf, that can be found...
-
changes for new wxwidgets library
-
renamed file to make compatible with windows
-
removed CMakeStuff from cvblob
-
- Added floorplan generation
-
Hello Nick, slam6d writes the initial transformation, intermediate steps and the...
1 >