NVAstro

This project aims to replicate the functions of DeepSkyStacker, speeding it up by offloading work to a CUDA-capable GPU whenever appropriate.

---

Functions

Below's the checklist of functions to be implemented, roughly in order of their application according to DSS.

• Alignment
  • Star detection
  • Offset & angle computation
  • Applying transformation
• Stacking
  • Background calibration
  • Flat frame calibration
  • Hot pixel detection & removal
  • [Low Priority] Removal of bad columns
  • [Low Priority] Entropy based dark frame subtraction
  • Stacking [WIP]
    • Creating master offset, dark, flat frames
    • Methods:
      • Average
      • Median

---

Project Structure

• .vscode VSCode workspace settings.
• build Built products & intermediates.
• src Source files.
  • cuda The cuda code used in nvastro.
  • main.cpp The nvastro application's entrypoint.
  • test_cuda.cu The test_cuda target's entrypoint. It's linked with codes in the cuda folder, allowing stuff there to be tested without going through the main app.
• misc.ipynb Random experiments.

---

Compilation

Compiling this project requires the CUDA toolkit, g++ compiler, LibRaw & ImageMagick libraries.

In the root folder of this project, run:

make nvastro

Or if you want to make and run the test_cuda target:

make test_cuda