Vision/db_scan

[dimembermatt]

Description

Pull request to merge vision/db_scan with dev. Working version of module DB_SCAN3.py and readme instructions in the main subdirectory.

API Changes

Adds module db_scan with functions db_scan::output_single (a colored mask of all identified objects) and db_scan::output_individual (a set of masks with each individual object). Internal functions including coloring, separating masks, and db_scan are restricted from the user.

Tests

Run tests with python test_dbscan.py <input> <radius> <density> <options>:

• input is the file to load and run dbscan on
• radius is the ball radius to use
• density is the density threshold for clustering
• options can be single, separate, both, or time.

Todos

Things that need to be done but aren't critical (that you promise to get done)

• [ ] Moar tests! (Although it works pretty well already, unlike DB_SCAN4...)
• [ ] Integration with the rest of the stack to check scalability
• [ ] Optimization tests with more image edge cases DB_SCAN_runtime_tests.xlsx
• [x] Documentation

Notes

Works as is, robustly, and as of this moment, with no noticeable or life-threatening flaws. Requires an input greyscale image to work properly.

[thetianshuhuang]

Please remove all older versions of your code; that's what git is for! The idea is that only the most recent copy is kept in any given version; to access older versions, instead of having ..._v1, _v2, etc, you can just pull up an older commit. I'm also not sure why there are so many files in the commit; you should avoid committing test results and old tests.

[thetianshuhuang]

I organized the folders (dbscan -> vision/dbscan) and added some optimizations. The profile function is copied in; this will probably be turned into a submodule or core module later since it's too useful to be kept to just dbscan.