Skip to content

Latest commit

 

History

History
57 lines (47 loc) · 2.74 KB

README.md

File metadata and controls

57 lines (47 loc) · 2.74 KB
Raw

Cluster Analysis of Trimmed Spectrograms (CATS)

DOI

CATS is a signal processing technique and framework for detecting and denoising sparse signals in the time-frequency domain. Particularly, very useful for processing earthquakes. This work is still in progress, and the package is under active development. Soon, here will be links to our papers/preprints.

Key features of CATS

  • Versatile. Any signals (not necessarily seismic) that are sparse in the time-frequency domain can be localized by CATS.
  • Flexible. Any time-frequency transform can be used as a base (STFT, CWT, ...). Fast detection with STFT or more accurate denoising with CWT.
  • Fast and accurate. Here will be links to our papers showing this.
  • Transparent and QC-friendly.
    • Minimum number of parameters which are easy to autotune.
    • Interpretable and visualizable workflow steps and parameters.
    • Collected cluster statistics can be used for custom post-processing and quality control (QC).

Installation

To install the package:

  1. Short way: pip install git+https://github.com/sgrubas/cats.git
  2. Other way:
    1. Clone repository: git clone https://github.com/sgrubas/cats.git
    2. Open the cats directory: cd cats
    3. Install: 1) pip install . or 2) pip install -e . (editable mode)

Dependencies

The package was tested on Python 3.9. See other dependencies in requirements.txt.

Tutorials

Demos:

Signal detection with CATSDetector

Signal denoising with CATSDenoiser and CATSDenoiserCWT

Citation

If you find CATS useful for your research, please cite this repository (soon there will be links to our papers):

@article{grubas2023cats,
  title = {Cluster Analysis of Trimmed Spectrograms (CATS)},
  author = {Serafim Grubas and Mirko van der Baan},
  journal = {GitHub},
  url = {https://github.com/sgrubas/cats},
  year = {2024},
  doi = {10.5281/zenodo.13830301},
}

Authors