MATLAB scatter plots with millions of points are slow and resource intensive. However, most of the points are not visible since they are hidden by other points. This code detects which points are hidden and remove them.
The used algorithm is particularly efficient and can handle millions of points:
- a pixel matrix is generated
- the points are circle occupying a given number of pixels
- the indices of the points are placed (in order) in the pixel matrix
- the points that do not appear in the pixel matrix will be invisible in the plot
- the invisible points are removed
In other words, this algorithm work as a virtual graphic buffer. The plot is precomputed and invisible elements are deleted.
This algorithm (o(n) complexity) features several advantages:
- no need to compute the distance between all the points
- the memory requirement is linearly proportional to the number of pixels
- the memory requirement is linearly proportional to the number of scatter points
- computational cost is linearly proportional to the number of scatter points
This code has been successfully tested with large datasets:
- this algorithm is vectorized and many points are treated together.
- the number of points (chunk size) processed in a step can be selected.
- 100'000'000 points can be simplified in several minutes
Look at the examples run_example.m. A dataset with random points is successfully simplified (by a factor of 40) without changing the scatter plot result.
- Tested with MATLAB R2018b.
- No toolboxes are required.
- Compatibility with GNU Octave not tested but probably easy to achieve.
Thomas Guillod - GitHub Profile
This project is licensed under the BSD License, see LICENSE.md.
Thomas Guillod (2022). scatter_simplify_matlab (https://github.com/otvam/scatter_simplify_matlab), GitHub. Retrieved .
MATLAB Release Compatibility
Platform CompatibilityWindows macOS Linux
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!Start Hunting!