star-ID-graph-match

Source Code of "An Accurate Star Identification Approach Based on Spectral Graph Matching for Attitude Measurement of Spacecraft"

@article{you2022accurate,
  title={An accurate star identification approach based on spectral graph matching for attitude measurement of spacecraft},
  author={You, Zhiyuan and Li, Junzheng and Zhang, Hongcheng and Yang, Bo and Le, Xinyi},
  journal={Complex & Intelligent Systems},
  volume={8},
  number={2},
  pages={1639--1652},
  year={2022}
}

Qucik Start

1. Convert SAO star catalog to txt format

   sh scripts/convert.sh

2. Simulate to generate star images

   sh scripts/simulate.sh

3. Generate database

   sh scripts/gen_database.sh

4. Search for star identification with noise

   sh scripts/search.sh #STD_POSITION_NOISE #NUM_LOST_STAR #NUM_FALSE_STAR

   For example, sh scripts/search.sh 3.0 1 0 means the std of positon noise is 3.0 pixel, the number of lost star is 1, and the number of false star is 0.

Parameter Selection

• Select the best radius of the neighbor circle, i.e., $r$ in the paper

  sh scripts/para_select/select_best_R_AD.sh

• Select the uncertainty of the MST sum, i.e., $\Delta w$ & $\Delta w^{max}$ in the paper

  sh scripts/para_select/select_best_eps_mst.sh

Explanation

• The directories utils/database & utils/search share some similar scripts, including

  • select_R_AD_helper.py, graph_helper.py

    Difference. The scripts under utils/database use Right Ascension & Declination to select the neighbor stars and generate graph, while the scripts under utils/search use Pixel Coordinates.

  • k_vector_helper.py

    Difference. The script under utils/database constructs K-Vector Range Search Model, while the script under utils/search searches using the constructed Search Model.

Simulator UI

• We provide a simulator UI to generate star images

  sh scripts/run_panel.sh