MMTO - Multi-spectral Astronomical Source Detection Tool

MMTO (Faezi et al.) is a multi-spectral photometric object detection and color extraction software, representing and processing on max-tree (Salembier et al.) data structure across multiple spectral bands, built on .

• Multi-band processing: Simultaneous analysis across multiple spectral bands
• Cross-band matching: Intelligent source association between bands using spatial and similarity metrics

Table of Contents

• Overview
• Installation
  • Dependencies
• Usage
  • Configuration
  • Tuned run
  • Command line arguments
• Citation
• Bibliography
• License

Overview

MMTO extends multiple max-trees by integrating semantically meaningful node partitions, derived from statistical tests, into a structured graph. This integration enables the exploration of correlations among cross-band emissions, enhancing segmentation accuracy.

MMTO processing pipeline workflow.

Installation

Dependencies

The dependencies are listed in the ./requirements directory.

python -m venv ./venvs/mmto
source ./venvs/mmto/bin/activate
pip install -r ./requirements/requirements_base.txt
pip install -r ./requirements/requirements_torch.txt || pip install -r ./requirements/requirements_torch_fallback.txt
pip install -U pip setuptools wheel scikit-build cmake ninja
pip install --no-build-isolation ./mmtolib

Usage

Tip

MMTO supports any number of bands to be processed. In this version, input images should have the same resolution (pixels) and sly coordinates.

Configuration

Configure the bands and processing parameters (see MTO2 documentation) in config.yaml file. Here is an instance of 4-band setting:

bands:
  
  band_1:
    file_path: "./1st.fits"
    background_mode: "const"
    move_factor: 5
    area_ratio: 0.90
    s_sigma: 4.0
    G_fit: false
    skip_reduction: true
  
  band_2:
    file_path: "./2nd.fits"
    background_mode: "morph"
    move_factor: 8
    area_ratio: 0.90
    s_sigma: 4.0
    G_fit: true
    skip_reduction: false
  
  band_3:
    file_path: "./3rd.fits"
    background_mode: "const"
    move_factor: 8
    area_ratio: 0.93
    s_sigma: 2.7
    G_fit: true
    skip_reduction: true
  
  band_4:
    file_path: "./4th.fits"
    background_mode: "const"
    move_factor: 3
    area_ratio: 0.93
    s_sigma: 3.13
    G_fit: false
    skip_reduction: true

Tuned run

python main.py --co_sim 0.9 --pix_dist 3.0 

Command line arguments

Option	Description	Type	Default	Range/Values
--pix_dist	Cross-band central distance (pixel)	float	3.00	≥ 0
--co_sim	Cross-band cosine similarity threshold	float	0.90	[0.0, 1.0)
-h, --help	Show the help message and exit	flag	-	-

Citation

If you use MMTO in your research, please cite the following paper:

@ARTICLE{10535192,
  author={Hashem Faezi, Mohammad and Peletier, Reynier and Wilkinson, Michael H. F.},
  journal={IEEE Access}, 
  title={Multi-Spectral Source-Segmentation Using Semantically-Informed Max-Trees}, 
  year={2024},
  volume={12},
  number={},
  pages={72288-72302},
  doi={10.1109/ACCESS.2024.3403309}
}

Bibliography

• Faezi M. H., Peletier R., & Wilkinson M. H. (2024). “Multi-Spectral Source-Segmentation Using Semantically-Informed Max-Trees”. In: IEEE Access 12, pp. 72288 - 72302. DOI: 10.1109/ACCESS.2024.3403309.
• Salembier P., Oliveras A., & Garrido L. (1998). “Antiextensive connected operators for image and sequence processing”. In: IEEE Transactions on Image Processing 7.4, pp. 555–570. DOI: 10.1109/83.663500.

License

This project is licensed under the MIT License - see the LICENSE file for details.