-
Notifications
You must be signed in to change notification settings - Fork 1
Meeting Minutes
Clare DuVal edited this page Jul 16, 2019
·
8 revisions
- More slight debugging
- Validate files, check
- Sanity check (unit test) with rotations and translations applied to randomly selected pcpds point cloud
- Add noise to all points in a selection and try to compare
- Expect to break
- Lose epsilon from left, gain epsilon from right
- How quickly the accuracy gets worse/better with the bigger the epsilon
- Only break up x and y linearly, NOT z
- Test against other lidar files
- Check how many points are in a diagram
- Check issue with multiple zeros (not more than one bottleneck should be zero)
- Make sure image is not 2D, if is reevaluate using Rips, if so height filtration
- Crust, power-crust
- Find 3D lidar data, test
- PyViz, "C born"?, PPTK**, some other data point visualization tool
- Focus on read/write/view windows
- Reorganize files in folders, group modules
- Remove pip install
- Encode dependencies in requirements.txt file
- Update by Wednesday, July 3rd to meet on Friday, 5th
- Percentage length of boxes, rectangular prism
- Shingling, moving window
- Boxed representations, splitting boxes down
- Focus on base code, then implement boxes
- Ignore edge cases for now
- Test exact, slight coordinate slide, then rotate
Start with smallest data set.
- Usable data
- Find Python Libraries for
- Anaconda
- Reading in .Las files, optimal with library capable of other transformations
- Numpy
- Generating persistence diagrams
- Dionysus
- Setup github pages
- Read las files into objects - transform into x, y, z
- Split the information into sections, grid with variable passing into the main
- Scale parameter, absolute parameter
- Overwriting the method to use, or use separate functions
- Process sections into objects (may do this earlier) - save this data for holding
- Persistence diagrams with x, y, z coordinate
- Store in the persistence diagram
- Store “PCPDS” object as json files