This toolkit was created to perform more efficient computing while processing blocks of aerial oblique imagery blocks eg. Leica CityMapper data.
Whole toolkit is just .py file collection, so there's no need to perform real installation. To make sure the toolkit works just install python requirements USING METASHAPE BUILT-IN INTERPRETER. If your not sure how to do it you can try to use configuration script placed in scripts directory. Just run it with any python on your PC.
All stages must be run sequentially one after another.
There are certain requirements that your project must apply to for toolkit to work properly:
- CRS - make sure you use some kind of projected cartesian coordinate system (eg. UTM).
- Rotation - currently there's no Roll-Pitch-Heading to Omega-Phi-Kappa conversion mechanism implemented inside our toolkit, so make sure to convert your project to Omega-Phi-Kappa.
- Camera calibration - our toolkit uses camera calibration for few calculations. Therefore, you need to specify
either:
- Focal length in pixels
- Focal length in mm + Physical pixel size
First step of processing is to create 2D footprint for every image in the block. Footprint is calculated based on initial EO (Reference) and camera calibration image matrix size and focal lenght (Tools/Camera calibration). Footprints will be placed in new shape group called "Footprints".
For better image alignment we need to split images into directions (Nadir, Front etc.). Currently image from every direction will be moved to new camera group, so if your chunk has important directories structure be ready to lose it...
When working with big datasets you can use this tool to perform two main tasks:
- Filter images by AOI - if there's other shape group except Footprints containing just one polygon feature images which footprints does not overlap with AOI polygon will be moved to new camera group.
- Build processing blocks - additionally, you can "cut" AOI polygon with line features and split your data to multiple chunks, that will be processed separately.
Images will be matched using built-in Agisoft Metashape algorithm, but pairs will be generated based on footprints overlap.
Based on previous image matching now algorithm will perform proper Bundle Adjustment. Alignment will be performed sequentially - first Oblique images, then Front, Left and so on...
This function will filter tie points based on its projections. It's recommended to make copy of aligned chunk first. Filtering step is optional. After execution make sure to optimize cameras again.