Preprocessing and 3D Mapping

The Preprocessing and 3D Mapping GUIs in Mavis are organized using a top-down approach.

3D Mapping only requires selecting the image workspace after Preprocessing and then pressing OK when using the defaults. Here all steps apart from generating the Tie-point Cloud are required and activated by default.

The Preprocessing section requires dataset specific settings to correctly georeference the image sets. This refers to the availability of log files and additional information. Georeferencing is a crucial step and should be done carefully.

The Preprocessing workflows are similar, but depending on existing geotags or the availability of an Ardpilot dataflash log file and a camera log file, different options are mandatory or optional. Image processing, like resizing and color enhancements, are optional as is quality control.

The following examples are the basic workflows under ideal conditions. In general the Georeferencing and projecting section will be the most relevant.

Preprocessing

Geotagged images

  1. Workspace and Setup: Click Browse to navigate to the workspace containing the images
  2. Image processing (optional): Click OK to resize all images to 25-50% for faster processing
  3. Georeferencing and projecting: Click OK
  • Note: qualitiy control cannot be used because information on pitch, roll and yaw are not extracted from the images tags.


Use an Ardupilot dataflash log for georeferencing

  1. Workspace and Setup: Click Browse to navigate to the workspace containing the images
  2. Image processing (optional): Click OK to resize to 25-50% for faster processing
  3. Log file analysis: Click Browse to open the corresponding Ardupilot .log file
  4. Georeferencing and projecting: Settings: set the number of images taken before arming
  5. Georeferencing and projecting: Click OK
  • Note: All images of the flight have to be placed in the workspace. Missing images cannot be handled. It is therefore important that the flight parameters (navigation speed, altitude agl., trigger distance) match the cameras capabilities.


Use an Ardupilot dataflash log and an additional KAPTURE camera .cam file or a TimeTag.cam log file generated by Mavis for georeferencing

  1. Workspace and Setup: Click Browse to navigate to the workspace containing the images
  2. Image processing (optional): Click OK to set all images to landscape mode and resize to 25-50% for faster processing if required
  3. Log file analysis: Click Browse to open the corresponding Ardupilot .log file
  4. Log file analysis: Click Browse to open the corresponding camera .cam log file or TimeTag.cam file
  5. Georeferencing and projecting: Settings: set the number of images taken before arming
  6. Georeferencing and projecting: Click OK
  • Note: Mavis automatically generates an TimeTags.cam file containing the relative shooting time of the images. This can be used in georeferencing if images are missing in case the camera was too slow to take every picture the autopilot triggered. If a KAPTURE .cam file is available it should be used instead because it has a precision of about 10ms. In contrast the TimeTag.cam files are based on the time tags of the images and thus have a precision of about 1s, which is not very accurate. However, in cases of missing images it helps to 'save' the dataset. If a TimeTag.cam file is used the GPS accuracy in 3D Mapping should be set to the navigation speed in m/s.

3D Mapping

1) Workspace: Click Browse to navigate to the workspace containing the images 2) Click OK

*Notes:

  • Proper preprocessing is mandatory
  • Tie-point cloud generation is optional
  • All other setting should remain on their defaults

results matching ""

    No results matching ""