The Ladybug Export dialog is quite similar to the ScanLook Point Cloud dialog. You select or define the project folder, make a few setting adjustments, select the output area(s), and finally export the panorama's or geo-reference the imagery for use in colorization.
The Project Folder is used to select or specify the Ladybug project folder. This can be the same as the ScanLook lidar folder or a sub-folder of the same.
The Output Folder generally defaults to a GeoRef sub-folder within the Project Folder or sub-folder containing the raw pgr files. It can be set manually to almost any location.
The INS Event File should be automatically detected and is a simple text file with or without a header. A little further down there is an example.
For a normal NovAtel INS this file is generally an exported file from NovAtel Inertial Explorer (and typically at 200Hz). If using NovAtel with an SBAS or RTK solution then the Convert4 program is used to export the INSPVA records as an asc (*.asc) file. If using an OxTS xNav system, then there will generally be a csv file. If using a VectorNav system, then there will be a gps file.
The STA File is a station file and will most likely be selected automatically. Again, it is possible to change this by a manual selection. In the case of multiple sensors (such as dual FARO's, etc.) it could possibly select the wrong station file so a little bit of caution needs to be exercised in reviewing this field. Be sure that the station file is correctly selected. The station file contains necessary timing information to synchronize the sensor to the INS.
The […] fields are file or folder browser options found in most applications.
The REM button is used to Remove the selected project from the project folder list. This is generally unnecessary as the most-recently-used list will discard the oldest projects eventually anyhow.
The Edit button invokes the system editor for the particular file currently selected.
Filter Distance. If left blank, no thinning takes place. If a value greater than 0 then the image data is thinned to that distance interval. This is an easy way to get rid of extra images that may be unnecessary at slower speeds and when sitting still.
Output File Size. By default this is 10000×5000. Alternative sizes can be chosen.
Filename Prefix. This is the text that is used to prefix all output image file names with the default being LB_.
Format. The format defaults to jpg (with a consequent Q-factor compression option) but other formats include bmp, png, tif, and h264.
If jpg is selected as the format, then the Q-factor fields are presented. The Best setting gives the least compression with the least loss of data whereas the Worst gives a maximum compression with some image loss.
Extra Commands. This field can be used to enter additional commands that are rarely used but sometimes necessary.
Here is a list of commands: -r NNN-NNN The frame range to process. The first frame is 0. Default setting is to process all the images. -g GPS_OUTPUT_PATH Output GPS file prefix. Default is ladybugGPSOutput -w NNNNxNNNN Output image size (widthxheight) in pixel. Default is 2048x1024. -t RENDER_TYPE Output image rendering type: pano - panoramic view (default) dome - dome view spherical - spherical view rectify-0 - rectified image (camera 0) rectify-1 - rectified image (camera 1) rectify-2 - rectified image (camera 2) rectify-3 - rectified image (camera 3) rectify-4 - rectified image (camera 4) rectify-5 - rectified image (camera 5) -f FORMAT Output image format: bmp - Windows BMP image jpg - JPEG image (default) tiff - TIFF image png - PNG image h264 - H.264 video -c COLOR_PROCESS Color processsing method: hq - High quality linear method (default) hq-gpu - High quality linear method edge - Edge sensing method near - Nearest neighbor method near-f - Nearest neighbor(fast) method rigorous - Rigorous method down4 - Downsample method mono - Monochrome method df - Directional filter method -b NNN Blending width in pixel. Default is 100. -v X.XX Falloff correction value. Default is 1.000000. -a true/false Enable falloff correction. true - Enable. false - Disable. Default is false. -s true/false Enable software rendering. true - Enable. false - Disable. Default is false. -k true/false Enable anti-aliasing. true - Enable. false - Disable. Default is false. -z true/false Enable stabilization. true - Enable. false - Disable. Default is false. -n N Stabilization paramnter - Number of templates. Default is 6. -m NNN Stabilization paramnter - Maximum search region. Default is 100. -d X.XX Stabilization paramnter - Decay rate. Default is 0.950000. -q XXX Field of view in degrees when RENDER_TYPE is "spherical". Default is 60.000000. -Q XXX JPEG quality setting when saving images as JPEG. -x XXX-YYY-ZZZ Euler rotation angle in degrees when RENDER_TYPE is "spherical". Default is 0.000000-0.000000-0.000000. -l CAL_FILE_PATH Path to calibration file to replace. -e BITRATE Bitrate in kbps for H.264 video output. Default is 4000.
Render Type. This field defaults to pano but also includes dome, spherical, and TopoDOT. TopoDOT simply exports each individual camera perspective (6 for the LB5).
Color Processing Method. The default option for color processing is hq. Here is a list of each option and briefly what it means.
hq - High quality linear interpolation. This algorithm provides similar results to Rigorous, but is up to 30 times faster. hq-gpu - High quality linear interpolation - This algorithm is the same ashq, but the color processing is performed on GPU. edge - Edge sensing demosaicing - This is the most accurate method that can keep up with the camera's frame rate. near - Nearest neighbour demosaicing. near-f - Nearest neighbour demosaicing (fast) - Faster, less accurate than nearest neighbor demosaicing. rigorous - Rigorous demosaicing - This provides the second best quality colour reproduction. This method is very processor intensive and may not keep up with the camera's frame rate. Best used for offline processing where accurate colour reproduction is required. down4 - Downsample4 mode - Color process to output a half width and half height image. This allows for faster previews and processing. This results in an output image that is 1/4 the size of the source image.
Event Offset. For some Ladybug systems there is an offset from the first image frame and the first event recorded. Normally the offset is 0 but sometimes we have seen it as +/- 1 or 2. If you change this value, you must hit Compute to update the files to apply the new offset.
The Q-Factor allows you to change the JPEG compression from Worst (super compressed, blocky) to Best (almost no compression).
The Recenter command allows you rotate about the Z axis to change the default orientation. In some cases users have the 0 camera slightly off center to the center of the image is not looking along the roadway. Changing this value from 0 to 360 allows for the view point to change.
Apply Fall-off Correction. This value defaults to 1. For details on how to use this parameter, check with the Pt Grey (now FLIR) website.
The boresight settings are always entered in meters (for XYZ) and decimal degrees (for Omega, Phi, Kappa). Currently you get these values from LIDAR USA after you finalize your Ladybug mount relative to the INS. The XYZ values can be measured directly but are normally determined during boresight calculation. Likewise, omega and phi will nominally be 0 with Kappa being between 0 and 360 degrees depending upon how it is mounted.
The XYZ offsets can be measured by hand in which, X is almost 0, Y is along the axis of the vehicle (forward towards the cab is -Y), and Z is the height above the center of the IMU with +Z being up. O and P will be close to 0 and K will be close to -90.
The Read INS Units command reads the INS data file and determines what units the trajectory was output in.
The Match INS Units option, if checked, simply eliminates any conversion of the trajectory from the initial state.
The units can be selected to change the trajectory from one measurement unit to another.
After entering the Project settings, Output settings, and proper Boresight values, then the geo-referencing information for each frame can be computed along with thinning the data to only those frames necessary.