# Welcome to the STARMATCH package
[](https://pypi.python.org/pypi/starmatch/) [](https://pypi.python.org/pypi/starmatch/) [](https://pypi.python.org/pypi/starmatch/) [](https://GitHub.com/lcx366/STARMATCH/graphs/contributors/) [](https://GitHub.com/lcx366/STARMATCH/graphs/commit-activity) [](https://github.com/lcx366/STARMATCH/blob/master/LICENSE) [](http://starmatch.readthedocs.io/?badge=latest) [](https://travis-ci.org/lcx366/starmatch)
STARMATCH is a Python package designed for **star map matching**, **camera calibration**, **astronomical positioning**, and **distortion correction**. This package is ideal for researchers, engineers, and enthusiasts working in fields such as optical astronomy, satellite tracking, and space object monitoring. It integrates advanced algorithms to achieve high accuracy in celestial positioning and image correction.
## đ Key Features
**1. Star Map Matching**
- Automatically match star maps and estimate the **initial camera pointing direction** without prior information about the cameraâs orientation or field of view.
- Calibrate and fine-tune the **camera center pointing** for precise alignment when an approximate pointing direction is available.
**2. Astronomical Positioning**
- Derive precise **celestial coordinates (RA, Dec)** for spatial objects using calibrated camera orientation.
**3. Apparent Magnitude Estimation**
- Estimate **apparent magnitudes** of spatial objects if brightness is given.
**4. Distortion Correction**
- Build **camera distortion-correction models** to simulate lens distortions. Supported distortion models include:
- **Standard Radial Distortion Model (SRDM)**
- **Division-mode Radial Distortion Model (DRDM)**
- **Tangential Distortion Model (TDM)**
- **BrownâConrady Distortion Model (BCDM)**
- Apply distortion-correction methods to ensure accurate astronomical positioning. Supported methods include:
- 2D polynomial fitting based on distortion models.
- Gaussian Process Regression.
- Piecewise-Affine: The transform is based on a Delaunay triangulation of the points to form a mesh. Each triangle is used to find a local affine transform.
## đ ď¸ How to Install
To install STARMATCH, simply use `pip` in your terminal:
```
pip install starmatch
pip install starmatch --upgrade # to upgrade a pre-existing installation
```
## đ How to Use
### Star Map matching for Astronomical Images
To perform star map matching using the STARMATCH package, you will need to prepare a set of star catalog files and indices in advance. This process can be efficiently handled using the starcatalogquery package, which can be installed via:
```bash
pip install starcatalogquery
```
The preparation involves multiple steps, including creating a **simplified star catalog**, generating **index files and databases**, and creating **h5-formatted hash files**(only for blind star map matching). For more details, refer to [STARQUERY](https://github.com/lcx366/STARQUERY).
**Step 1: Load the Simplified Star Catalog**
```python
>>> from starcatalogquery import StarCatalog
>>> # Load a star catalog directory containing tiles of size 1.83 degrees.
>>> # The catalog includes stars brighter than magnitude 13.0, with proper motion corrections for epoch 2019.5.
>>> dir_from_simplified = 'starcatalogs/simplified/at-hyg24/mag13.0/epoch2019.5/'
>>> sc_simplified = StarCatalog.load(dir_from_simplified)
```
**Step 2: Extract Sources from Images or Files**
The Python package **starextractor** can be used for source extraction:
```bash
pip install starextractor
```
For more details, see [STAREXTRACTOR](https://github.com/lcx366/STAREXTRACTOR).
To simplify, hereâs how to load a test file (test.txt) containing pixel coordinates and grayscale values of extracted sources:
```python
>>> import numpy as np
>>> data = np.loadtxt('obs/test.txt')
>>> # Translate the origin of coordinates to the center of the image
>>> x = data[:, 0] - 512 # The resolution is (1024,1024)
>>> y = 512 - data[:, 1]
>>> xy = np.stack((x, y), axis=1)
>>> flux = data[:, 2]
```
đĄ **Note**: Ensure pixel coordinates are sorted by grayscale values in descending order. If not, sort them before proceeding unless you can guarantee these are the brightest stars in the image.
**Step 3: Load Sources and Camera Parameters into StarMatch**
Two methods for calculating geometric invariants are provided:
- **Triangles** (using three stars)
- **Quads** (using four stars)
```python
>>> from starmatch import StarMatch
>>> # Configure camera parameters: FOV[deg], pixel width[deg], and resolution
>>> camera_params = {'fov':(2,2),'pixel_width':0.002,'res':(1024,1024)} # The resolution is mandatory
>>> mode_invariants = 'quads' # Options: 'triangles' or 'quads'
>>> sources = StarMatch.from_sources(xy,camera_params,flux_raw=flux,mode_invariants=mode_invariants) # No distortion correction is applied
```
**Step 4: Estimate the Cameraâs Center Pointing**
**This step is optional if the approximate pointing is known.**
For blind matching, it is recommended to use the **Quads** matching method.
```python
>>> sc_hashed = sc_simplified.read_h5_hashes() # Read the h5-formatted Hash File
>>> # Estimate the initial camera pointing
>>> fp_radec,pixel_width_estimate,fov_estimate = sources.center_pointing(sc_hashed)
>>> print(fp_radec,pixel_width_estimate,fov_estimate)
```
**Step 5: Star Map Matching with Known Pointing**
If the camera orientation is approximately known, use the **Triangles** matching method.
```python
>>> fp_radec = [141.8,-2] # Approximate pointing [Ra,Dec] in degrees
>>> astrometry_corrections = {
't': '2019-02-26T20:11:14.347',
'proper-motion': None,
'aberration': (0.5595, -1.1778, 7.5032),
'parallax': None
}
>>> # Perform alignment with distortion correction
>>> sources.align(fp_radec,sc_simplified,astrometry_corrections=astrometry_corrections,distortion_calibrate='gpr')
>>> print(sources)
>>> # Display results
>>> print(sources.affined_results)
>>> print(sources.matched_results)
>>> print(sources.calibrated_results)
```
**About Astrometric corrections:**
astrometry_corrections -> [dict] Dictionary specifying the types of corrections to apply.
- 't' -> [str] Observation time in UTC, such as '2019-02-26T20:11:14.347'.
It specifies the time at which corrections are applied.
- 'proper-motion' -> [None] If present, apply proper motion correction.
This term corrects for the motion of stars across the sky due to their velocities.
- 'aberration' -> [tuple] Aberration correction parameters. Observer's velocity relative to Earth's center (vx, vy, vz) in km/s.
This term corrects for the apparent shift in star positions due to the motion of the observer relative to the Solar System Barycenter.
- 'parallax' -> [None] If present, apply parallax correction.
This term corrects for the apparent shift in star positions due to the change in observer's viewpoint as the Earth orbits the Sun.
- 'deflection' -> [None] If present, apply light deflection correction.
This term corrects for the bending of light from stars due to the gravitational field of the Sun, based on general relativity.
**About Distortion Calibration:**
- 'gpr': Nonparametric Gaussian Process Regression(GPR).
- 'piecewise-affine': The transform is based on a Delaunay triangulation of the points to form a mesh. Each triangle is used to find a local affine transform.
- 'polynomial': 2D polynomial transformation with the following form
$X = \sum_{j=0}^2 \sum_{i=0}^j a_{ji} x^{j - i} y^i$
$Y = \sum_{j=0}^2 \sum_{i=0}^j b_{ji} x^{j - i} y^i$
**Explanation of Results:**
- `affined_results`: Initial alignment results.
- `matched_results`: Enhanced results using more sources.
- `calibrated_results`: Final results after applying distortion correction.
Each result includes:
- xy: Computed pixel coordinates of sources.
- xy_res: Residuals of pixel coordinates.
- xy_rms: RMS of of pixel coordinates.
- mag_res: Residuals of magnitudes of sources.
- mag_rms: RMS of magnitudes.
- C: Magnitudes constant.
- C_sigma: Uncertainty of magnitudes constant.
- catalog_df: DataFrame of matched stars.
- _description: Results description.
- pixels_camera_match: Pixel coordinates of sources.
- radec_res: Residuals of celestial coordinates.
- radec_rms: RMS of celestial coordinates.
**Step 6: Calibrate the Cameraâs Orientation**
```python
>>> sources.fp_calibrate()
>>> print(sources.fp_radec_calibrated)
```
**Step 7: Visualize the Distortion Map**
Generate a vector plot showing the distortion at various points:
```python
>>> sources.show_distortion('vector')
```
<p align="middle">
<img src="readme_figs/output_70_1.png" width="500" />
</p>
To visualize the distortion in the x and y directions using a contour plot:
```python
>>> sources.show_distortion('contourf')
```
<p align="middle">
<img src="readme_figs/output_73_1.png" width="800" />
</p>
### Astrometric Positioning and Magnitude Estimation
Estimate celestial coordinates and apparent magnitudes of spatial objects.
```python
>>> # Ensure the coordinates origin of targets are translated to the center of the image as done for stars
>>> x_target,y_target = 125.717 - 512,512 - 397.795
>>> xy_target = [x_target,y_target]
>>> flux_target = 3003.62
>>> radec,mag_affine,mag_match = sources.apply(xy_target,flux_target)
```
**đĄ Note**: The celestial coordinates estimated above represent the apparent position, not the true position. When the relative velocity between the observer and the space target is significant, corrections for **aberration** and **light-time effects** are essential. The apparent direction at time t corresponds to the true direction at time tâĎ (where Ď is the light-time offset). Applying this correction helps account for deviations that can reach several arcseconds.
### Visualize Matched Stars on the Image
To display matched stars directly on the original astronomical image:
```python
>>> from starextractor import parse_image
>>> # Load the image (supports FITS or BMP formats)
>>> image_file = 'obs/fits/img_00000.fits' # or 'obs/bmp/img_00000.bmp'
>>> image_data = parse_image(image_file)
>>> # Extract sources from the image
>>> sources = image_data.find_source()
>>> # Configure camera parameters
>>> camera_params = {'fov': (8, 8), 'pixel_width': 0.01, 'res': (1024, 1024)}
>>> sources = StarMatch.from_sources(sources.xy, camera_params, flux_raw=sources.brightness)
>>> # Perform star map alignment using a known approximate orientation
>>> fp_radec = [201, 31] # Example pointing direction [RA, Dec] in degrees
>>> sources.align(fp_radec, sc_simplified)
>>> # Display matched stars on the image
>>> sources.show_starmatch(image_data.image_raw, image_data._offset)
```
<p align="middle">
<img src="readme_figs/output_52_0.png" width="500" />
</p>
The stars marked in the image should align with entries in `catalog_df`:
```python
print(sources.matched_results.catalog_df)
```
<p align="middle">
<img src="readme_figs/catalog_df.png" width="700" />
</p>
### Troubleshooting on Star Map Matching
If star map matching fails, consider the following:
1. **Image Coordinates**: Ensure that the origin of the image coordinates is correctly set. The origin may differ (e.g., upper-left corner vs. lower-left corner). Adjust the pixel coordinates of the sources accordingly.
2. **Field of View**: Select an appropriate star catalog based on the field of view:
- **Bright star catalog**: Suitable for a large field of view.
- **Faint star catalog**: Suitable for a smaller field of view.
3. **Geometric Invariants**: When constructing geometric invariants, the default number of stars used in the **kd-Tree nearest neighbor search** is 9. Increasing this number may **improve the success rate of matching**.
4. **Blind Matching**: For small fields of view, consider increasing the **HEALPix level** to enhance the success rate of blind matching.
### Geometric Distortion Model
Available distortion models include Standard Radial Distortion Model(SRDM), Division-mode Radial Distortion Model(DRDM), Tangential Distortion Model(also known as the de-centering distortion), and Brown-Conrady Distortion Model(BCDM).
Considering that the distortion model involves the power of the distance from the distortion center to pixels, the length scale is introduced here. Associaated with the length scale, the normalized model coefficients is also necessary.
#### Example: Standard Radial Distortion Model(SRDM)
The Standard Radial Distortion Model(SRDM) is defined by the distortion center and the distortion coefficients.
- The 3rd order polynomial in standard form: $r_u = r_d + k_1 r_d^3$
- The 5th order polynomial in standard form: $r_u = r_d + k_1 r_d^3 + k_2 r_d^5$
- The 5th order polynomial in all form: $r_u = r_d + k_1 r_d^2 + k_2 r_d^3 + k_3 r_d^4 + k_4 r_d^5$,
where $r_d$ is the distance between the distorted pixel coordinates and the distortion center, $r_u$ is the distance between the distortion-corrected pixel coordinates and the distortion center.
1. The 3rd order polynomial in standard form only works well for small amounts of distortion.
2. The 5th order polynomial produce more accurate results, especially for âwaveâ or âmustacheâ distortion, which might resemble barrel near the center of the image and pincushion near the corners (or vice-versa).
3. The 5th order all form use all coefficients up to the maximum instead of alternate coefficients (odd or even-only). We have not observed much advantage to these settings.
4. Higher order polynomials (7th order or higher) should be used with great care because results can become unstable, especially at the outer parts of the image. The image used to calculate the coefficients should have valid corner points near the edge of the image and there should be sufficient rows or columns.
Basic formulas are as follows:
$x_u - x_c = r_u \cos(\theta) = r_u (x_d - x_c)/r_d = ... $
$y_u - y_c = r_u \sin(\theta) = r_u (y_d - y_c)/r_d = ... $
where $(x_c,y_c)$ is the pixel coordinates of the distortion center.
For more details, please refer to
1. [Distortion_(optics)](https://en.wikipedia.org/wiki/Distortion_(optics))
2. [imatest distortion-methods](https://www.imatest.com/docs/distortion-methods-and-modules/)
3. [imatest distortion-models](https://www.imatest.com/support/docs/pre-5-2/geometric-calibration-deprecated/distortion-models/)
##### Construct a Standard Radial Distortion Model
```python
>>> from starmatch.classes import Distortion
>>> model = 'RadialStandard' # Type of distortion model
>>> coeffs = [-1e-4,1e-4] # Coefficients of 5th order SRDM in form of [k1,k2]
>>> dc = [0.1,0.1] # Pixel coordinates of the distortion center
>>> distortion_scale = 128 # The length scale of the distortion model, that is, the number of pixels per unit length
>>> distortion = Distortion(model,coeffs,dc,distortion_scale) # Establish a distortion model
```
##### Compute the distortion-corrected pixel coordinates
```python
>>> pixels_xy = [367,125]
>>> pixels_XY = distortion.apply(pixels_xy)
>>> print(pixels_XY)
>>> # [[369.21610677 125.70199655]]
>>> #Calculate the distortion-corrected pixel coordinates at normalized scale
>>> #pixels_xy = [[2.8671875, 0.9765625], [1.109375, -0.875]]
>>> #pixel_scale = 128
>>> #pixels_XY = distortion.apply(pixels_xy,pixel_scale)
>>> #print(pixels_XY)
```
##### Sketch the distortion
Vector plot of the distortion:
```python
>>> xlim,ylim = 512,512
>>> distortion.sketchmap(xlim,ylim)
>>> #For normalized pixel coordinates
>>> #xlim,ylim = 4,4
>>> #pixel_scale = 128
>>> #distortion.sketchmap(xlim,ylim,pixel_scale)
```
<p align="middle">
<img src="readme_figs/output_8_1.png" width="500" />
</p>
Contour plot of the distortion in x and y components respectively:
```python
>>> distortion.sketchmap(xlim,ylim,mode='contourf')
```
<p align="middle">
<img src="readme_figs/output_8_1_contourf.png" width="800" />
</p>
#### Example: Division-mode Radial Distortion Model(DRDM)
The Division-mode Radial Distortion Model defined by the distortion center and the distortion coefficients.
- The 2nd order in division form: $r_u = r_d /(1+ k_1 r_d^2)$
- The 4th order in division form: $r_u = r_d /(1+ k_1 r_d^2 + k_2 r_d^4)$
Basic formulas for DRDM are same as that for SRDM.
##### Construct a Division-mode Radial Distortion Model
```python
>>> from starmatch.classes import Distortion
>>> model = 'RadialDivision' # Type of distortion model
>>> coeffs = [-1e-4,1e-4] # Coefficients of 4th order DRDM in form of [k1,k2]
>>> dc = [0.1,0.1] # Pixel coordinates of the distortion center
>>> distortion_scale = 128 # The length scale of the distortion model, that is, the number of pixels per unit length
>>> distortion = Distortion(model,coeffs,dc,distortion_scale) # Establish a distortion model
```
##### Compute the distortion-corrected pixel coordinates
```python
>>> pixels_xy = [367,125]
>>> pixels_XY = distortion.apply(pixels_xy)
>>> print(pixels_XY)
>>> # [[364.79767243 124.30236829]]
>>> #Calculate the distortion-corrected pixel coordinates at normalized scale
>>> #pixels_xy = [[2.8671875, 0.9765625], [1.109375, -0.875]]
>>> #pixel_scale = 128
>>> #pixels_XY = distortion.apply(pixels_xy,pixel_scale)
>>> #print(pixels_XY)
```
##### Sketch the vector plot of distortion
```python
>>> xlim,ylim = 512,512
>>> distortion.sketchmap(xlim,ylim)
>>> #For normalized pixel coordinates
>>> #xlim,ylim = 4,4
>>> #pixel_scale = 128
>>> #distortion.sketchmap(xlim,ylim,pixel_scale)
```
<p align="middle">
<img src="readme_figs/output_15_1.png" width="500" />
</p>
#### Example: Tangential Distortion Model(TDM)
The Tangential Distortion Model(also known as the de-centering distortion) is defined by the distortion center and the distortion coefficients.
Basic formulas are as follows:
$x_u = x_d + (P_1 (r_d^2 + 2 (x_d - x_c)^2) + 2 P_2 (x_d-x_c) (y_d-y_c)) (1 + P_3 r_d^2 + P_4 r_d^4 + ...)$
$y_u = y_d + (P_2 (r_d^2 + 2 (y_d - y_c)^2) + 2 P_1 (x_d-x_c) (y_d-y_c)) (1 + P_3 r_d^2 + P_4 r_d^4 + ...)$
##### Construct a Tangential Distortion Model
```python
>>> from starmatch.classes import Distortion
>>> model = 'Tangential' # Type of distortion model
>>> coeffs = [-1e-4,1e-4] # Coefficients of 2nd order SRDM in form of [P1,P2]
>>> dc = [0.1,0.1] # Pixel coordinates of the distortion center
>>> distortion_scale = 128 # The length scale of the distortion model, that is, the number of pixels per unit length
>>> distortion = Distortion(model,coeffs,dc,distortion_scale) # Establish a distortion model
```
##### Compute the distortion-corrected pixel coordinates
```python
>>> pixels_xy = [367,125]
>>> pixels_XY = distortion.apply(pixels_xy)
>>> print(pixels_XY)
>>> # [[366.75821931 125.06542319]]
>>> #Calculate the distortion-corrected pixel coordinates at normalized scale
>>> #pixels_xy = [[2.8671875, 0.9765625], [1.109375, -0.875]]
>>> #pixel_scale = 128
>>> #pixels_XY = distortion.apply(pixels_xy,pixel_scale)
>>> #print(pixels_XY)
```
##### Sketch the vector plot of distortion
```python
>>> xlim,ylim = 512,512
>>> distortion.sketchmap(xlim,ylim)
>>> #For normalized pixel coordinates
>>> #xlim,ylim = 4,4
>>> #pixel_scale = 128
>>> #distortion.sketchmap(xlim,ylim,pixel_scale)
```
<p align="middle">
<img src="readme_figs/output_22_1.png" width="500" />
</p>
#### Example: BrownâConrady Distortion Model(BCDM)
The BrownâConrady model corrects both the radial distortion and the tangential distortion caused by physical elements in a lens not being perfectly aligned.
Basic formulas are as follows:
$x_u = x_d + (x_d - x_c) (K_1 r_d^2 + K_2 r_d^4 + ...) + (P_1 (r_d^2 + 2 (x_d-x_c)^2) + 2 P_2 (x_d-x_c) (y_d-y_c)) (1 + P_3 r_d^2 + P_4 r_d^4 + ...)$
$y_u = y_d + (y_d - x_c) (K_1 r_d^2 + K_2 r_d^4 + ...) + (P_2 (r_d^2 + 2 (y_d-y_c)^2) + 2 P_1 (x_d-x_c) (y_d-y_c)) (1 + P_3 r_d^2 + P_4 r_d^4 + ...)$
##### Construct a BrownâConrady Distortion Model
```python
>>> from starmatch.classes import Distortion
>>> model = 'BrownâConrady' # Type of distortion model
>>> coeffs = [[-1e-4,1e-4],[1e-3,1e-3,1e-4,1e-5]] # Coefficients of BrownâConrady distortion model in form of [[coeffs_radial],[coeffs_tangential]]
>>> dc = [0.1,0.1] # Pixel coordinates of the distortion center
>>> distortion_scale = 128 # The length scale of the distortion model, that is, the number of pixels per unit length
>>> distortion = Distortion(model,coeffs,dc,distortion_scale) # Establish a distortion model
```
### Compute the distortion-corrected pixel coordinates
```python
>>> pixels_xy = [367,125]
>>> pixels_XY = distortion.apply(pixels_xy)
>>> print(pixels_XY)
>>> # [[372.88150908 127.60108593]]
>>> #Calculate the distortion-corrected pixel coordinates at normalized scale
>>> #pixels_xy = [[2.8671875, 0.9765625], [1.109375, -0.875]]
>>> #pixel_scale = 128
>>> #pixels_XY = distortion.apply(pixels_xy,pixel_scale)
>>> #print(pixels_XY)
```
##### Sketch the vector plot of distortion
```python
>>> xlim,ylim = 512,512
>>> distortion.sketchmap(xlim,ylim)
>>> #For normalized pixel coordinates
>>> #xlim,ylim = 4,4
>>> #pixel_scale = 128
>>> #distortion.sketchmap(xlim,ylim,pixel_scale)
```
<p align="middle">
<img src="readme_figs/output_29_1.png" width="500" />
</p>
## Change log
- **0.2.7 â Nov 19, 2024**
- Previously, the star map matching process was based on two iterations. An additional matching iteration has been introduced, significantly improving the accuracy of star map matching.
- **0.2.6 â Nov 18, 2024**
- Implemented different pixel tolerances for initial and secondary matching stages, improving the success rate for both star map matching and blind matching.
- **0.2.5 â Oct 30, 2024**
- Improved outliers identification in star map matching with the method of LOWESS (Locally Weighted Scatterplot Smoothing).
- **0.2.4 â Sep 29, 2024**
- In order to fit the distortion model, it is necessary to select as many stars as possible in the field of view. However, considering the computational speed budget, the selection of all stars is modified to select evenly distributed bright stars.
- The RLM(Robust Linear Model) is used instead of ordinary least squares to estimate the magnitude constant C and its uncertainty
based on observed fluxes and apparent magnitudes.
- **0.2.3 â Sep 03, 2024**
- Improved WCS transform.
- Optimized threshold parameters.
- **0.2.2 â Aug 27, 2024**
- By adding a secondary test, the error rate of blind star map matching is significantly reduced.
- **0.2.1 â Jul 09, 2024**
- Added two methods for outliers recognition in star map mismatching ahead of distortion calibration:
- 'lowess': Identifies outliers with the method of LOWESS (Locally Weighted Scatterplot Smoothing). LOWESS uses a weighted **linear regression** by default.
- 'iqr': Identifies outliers with the method of Interquartile Range (IQR).
- **0.2.0 â Jul 07, 2024**
- Compatible with astronomical corrections, including proper motion, annual parallax, aberration, and light deflection.
- Added two distortion estimation methods:
- 'piecewise-affine': The transform is based on a Delaunay triangulation of the points to form a mesh. Each triangle is used to find a local affine transform.
- 'polynomial': 2D polynomial transformation.
- Implemented a star map blind matching algorithm for astronomical images, which can adapt to the field of view of images from tens of degrees to tens of arc minutes.
- **0.1.4 â Sep 23, 2023**
- Fixed an issue where star chart matching failed after performing a blind match.
- **0.1.3 â Sep 05, 2023**
- Added contour plot for distortion models.
- Minor bugs fixed.
- **0.1.2 â Jul 23, 2023**
- Simplified the use of blind matching.
- Added pixel width and field of view estimates for blind matching.
- **0.1.1 â Jun 16, 2023**
- The ***starmatch*** package was released.
## Next release
- Find the inverse transformation of distortion models
## Reference
- [Distortion_(optics)](https://en.wikipedia.org/wiki/Distortion_(optics))
- [imatest distortion-methods](https://www.imatest.com/docs/distortion-methods-and-modules/)
- [imatest distortion-models](https://www.imatest.com/support/docs/pre-5-2/geometric-calibration-deprecated/distortion-models/)
- [Astroalign](https://astroalign.quatrope.org/en/latest/index.html)
Raw data
{
"_id": null,
"home_page": "https://github.com/lcx366/STARMATCH",
"name": "starmatch",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.10",
"maintainer_email": null,
"keywords": "Star Map Matching, Astrometric positioning, Distortion Calibration, Astronomical Correction",
"author": "Chunxiao Li",
"author_email": "lcx366@126.com",
"download_url": null,
"platform": null,
"description": "# Welcome to the STARMATCH package\n\n[](https://pypi.python.org/pypi/starmatch/) [](https://pypi.python.org/pypi/starmatch/) [](https://pypi.python.org/pypi/starmatch/) [](https://GitHub.com/lcx366/STARMATCH/graphs/contributors/) [](https://GitHub.com/lcx366/STARMATCH/graphs/commit-activity) [](https://github.com/lcx366/STARMATCH/blob/master/LICENSE) [](http://starmatch.readthedocs.io/?badge=latest) [](https://travis-ci.org/lcx366/starmatch)\n\nSTARMATCH is a Python package designed for **star map matching**, **camera calibration**, **astronomical positioning**, and **distortion correction**. This package is ideal for researchers, engineers, and enthusiasts working in fields such as optical astronomy, satellite tracking, and space object monitoring. It integrates advanced algorithms to achieve high accuracy in celestial positioning and image correction.\n\n## \ud83d\ude80 Key Features\n\n**1. Star Map Matching**\n\n- Automatically match star maps and estimate the **initial camera pointing direction** without prior information about the camera\u2019s orientation or field of view.\n\n- Calibrate and fine-tune the **camera center pointing** for precise alignment when an approximate pointing direction is available.\n\n**2. Astronomical Positioning**\n\n- Derive precise **celestial coordinates (RA, Dec)** for spatial objects using calibrated camera orientation.\n\n**3. Apparent Magnitude Estimation**\n\n- Estimate **apparent magnitudes** of spatial objects if brightness is given.\n\n**4. Distortion Correction**\n\n- Build **camera distortion-correction models** to simulate lens distortions. Supported distortion models include:\n \n - **Standard Radial Distortion Model (SRDM)**\n \n - **Division-mode Radial Distortion Model (DRDM)**\n \n - **Tangential Distortion Model (TDM)**\n \n - **Brown\u2013Conrady Distortion Model (BCDM)**\n\n- Apply distortion-correction methods to ensure accurate astronomical positioning. Supported methods include:\n \n - 2D polynomial fitting based on distortion models.\n \n - Gaussian Process Regression.\n \n - Piecewise-Affine: The transform is based on a Delaunay triangulation of the points to form a mesh. Each triangle is used to find a local affine transform.\n\n## \ud83d\udee0\ufe0f How to Install\n\nTo install STARMATCH, simply use `pip` in your terminal:\n\n```\npip install starmatch\npip install starmatch --upgrade # to upgrade a pre-existing installation\n```\n\n## \ud83d\udcda How to Use\n\n### Star Map matching for Astronomical Images\n\nTo perform star map matching using the STARMATCH package, you will need to prepare a set of star catalog files and indices in advance. This process can be efficiently handled using the starcatalogquery package, which can be installed via: \n\n```bash\npip install starcatalogquery\n```\n\nThe preparation involves multiple steps, including creating a **simplified star catalog**, generating **index files and databases**, and creating **h5-formatted hash files**(only for blind star map matching). For more details, refer to [STARQUERY](https://github.com/lcx366/STARQUERY).\n\n**Step 1: Load the Simplified Star Catalog**\n\n```python\n>>> from starcatalogquery import StarCatalog\n\n>>> # Load a star catalog directory containing tiles of size 1.83 degrees.\n>>> # The catalog includes stars brighter than magnitude 13.0, with proper motion corrections for epoch 2019.5.\n>>> dir_from_simplified = 'starcatalogs/simplified/at-hyg24/mag13.0/epoch2019.5/'\n>>> sc_simplified = StarCatalog.load(dir_from_simplified)\n```\n\n**Step 2: Extract Sources from Images or Files**\n\nThe Python package **starextractor** can be used for source extraction:\n\n```bash\npip install starextractor\n```\n\nFor more details, see [STAREXTRACTOR](https://github.com/lcx366/STAREXTRACTOR).\n\nTo simplify, here\u2019s how to load a test file (test.txt) containing pixel coordinates and grayscale values of extracted sources:\n\n```python\n>>> import numpy as np\n>>> data = np.loadtxt('obs/test.txt')\n>>> # Translate the origin of coordinates to the center of the image\n>>> x = data[:, 0] - 512 # The resolution is (1024,1024)\n>>> y = 512 - data[:, 1]\n>>> xy = np.stack((x, y), axis=1)\n>>> flux = data[:, 2]\n```\n\n\ud83d\udca1 **Note**: Ensure pixel coordinates are sorted by grayscale values in descending order. If not, sort them before proceeding unless you can guarantee these are the brightest stars in the image.\n\n**Step 3: Load Sources and Camera Parameters into StarMatch**\n\nTwo methods for calculating geometric invariants are provided:\n\n- **Triangles** (using three stars)\n\n- **Quads** (using four stars)\n\n```python\n>>> from starmatch import StarMatch\n\n>>> # Configure camera parameters: FOV[deg], pixel width[deg], and resolution\n>>> camera_params = {'fov':(2,2),'pixel_width':0.002,'res':(1024,1024)} # The resolution is mandatory\n>>> mode_invariants = 'quads' # Options: 'triangles' or 'quads'\n>>> sources = StarMatch.from_sources(xy,camera_params,flux_raw=flux,mode_invariants=mode_invariants) # No distortion correction is applied\n```\n\n**Step 4: Estimate the Camera\u2019s Center Pointing**\n\n**This step is optional if the approximate pointing is known.**\n\nFor blind matching, it is recommended to use the **Quads** matching method.\n\n```python\n>>> sc_hashed = sc_simplified.read_h5_hashes() # Read the h5-formatted Hash File\n>>> # Estimate the initial camera pointing\n>>> fp_radec,pixel_width_estimate,fov_estimate = sources.center_pointing(sc_hashed)\n>>> print(fp_radec,pixel_width_estimate,fov_estimate) \n```\n\n**Step 5: Star Map Matching with Known Pointing**\n\nIf the camera orientation is approximately known, use the **Triangles** matching method.\n\n```python\n>>> fp_radec = [141.8,-2] # Approximate pointing [Ra,Dec] in degrees\n>>> astrometry_corrections = {\n 't': '2019-02-26T20:11:14.347',\n 'proper-motion': None,\n 'aberration': (0.5595, -1.1778, 7.5032),\n 'parallax': None\n }\n>>> # Perform alignment with distortion correction\n>>> sources.align(fp_radec,sc_simplified,astrometry_corrections=astrometry_corrections,distortion_calibrate='gpr')\n>>> print(sources)\n\n>>> # Display results\n>>> print(sources.affined_results)\n>>> print(sources.matched_results)\n>>> print(sources.calibrated_results)\n```\n\n**About Astrometric corrections:**\n\nastrometry_corrections -> [dict] Dictionary specifying the types of corrections to apply.\n\n- 't' -> [str] Observation time in UTC, such as '2019-02-26T20:11:14.347'.\n It specifies the time at which corrections are applied.\n- 'proper-motion' -> [None] If present, apply proper motion correction.\n This term corrects for the motion of stars across the sky due to their velocities.\n- 'aberration' -> [tuple] Aberration correction parameters. Observer's velocity relative to Earth's center (vx, vy, vz) in km/s.\n This term corrects for the apparent shift in star positions due to the motion of the observer relative to the Solar System Barycenter.\n- 'parallax' -> [None] If present, apply parallax correction.\n This term corrects for the apparent shift in star positions due to the change in observer's viewpoint as the Earth orbits the Sun.\n- 'deflection' -> [None] If present, apply light deflection correction.\n This term corrects for the bending of light from stars due to the gravitational field of the Sun, based on general relativity.\n\n**About Distortion Calibration:**\n\n- 'gpr': Nonparametric Gaussian Process Regression(GPR).\n\n- 'piecewise-affine': The transform is based on a Delaunay triangulation of the points to form a mesh. Each triangle is used to find a local affine transform.\n\n- 'polynomial': 2D polynomial transformation with the following form\n \n $X = \\sum_{j=0}^2 \\sum_{i=0}^j a_{ji} x^{j - i} y^i$\n \n $Y = \\sum_{j=0}^2 \\sum_{i=0}^j b_{ji} x^{j - i} y^i$\n\n**Explanation of Results:**\n\n- `affined_results`: Initial alignment results.\n\n- `matched_results`: Enhanced results using more sources.\n\n- `calibrated_results`: Final results after applying distortion correction.\n\nEach result includes:\n\n- xy: Computed pixel coordinates of sources. \n\n- xy_res: Residuals of pixel coordinates. \n\n- xy_rms: RMS of of pixel coordinates. \n\n- mag_res: Residuals of magnitudes of sources. \n\n- mag_rms: RMS of magnitudes. \n\n- C: Magnitudes constant. \n\n- C_sigma: Uncertainty of magnitudes constant. \n\n- catalog_df: DataFrame of matched stars. \n\n- _description: Results description. \n\n- pixels_camera_match: Pixel coordinates of sources. \n\n- radec_res: Residuals of celestial coordinates. \n\n- radec_rms: RMS of celestial coordinates.\n\n**Step 6: Calibrate the Camera\u2019s Orientation**\n\n```python\n>>> sources.fp_calibrate()\n>>> print(sources.fp_radec_calibrated)\n```\n\n**Step 7: Visualize the Distortion Map**\n\nGenerate a vector plot showing the distortion at various points:\n\n```python\n>>> sources.show_distortion('vector')\n```\n\n<p align=\"middle\">\n <img src=\"readme_figs/output_70_1.png\" width=\"500\" />\n</p>\n\nTo visualize the distortion in the x and y directions using a contour plot:\n\n```python\n>>> sources.show_distortion('contourf')\n```\n\n<p align=\"middle\">\n <img src=\"readme_figs/output_73_1.png\" width=\"800\" />\n</p>\n\n### Astrometric Positioning and Magnitude Estimation\n\nEstimate celestial coordinates and apparent magnitudes of spatial objects.\n\n```python\n>>> # Ensure the coordinates origin of targets are translated to the center of the image as done for stars\n>>> x_target,y_target = 125.717 - 512,512 - 397.795\n>>> xy_target = [x_target,y_target]\n>>> flux_target = 3003.62\n>>> radec,mag_affine,mag_match = sources.apply(xy_target,flux_target)\n```\n\n**\ud83d\udca1 Note**: The celestial coordinates estimated above represent the apparent position, not the true position. When the relative velocity between the observer and the space target is significant, corrections for **aberration** and **light-time effects** are essential. The apparent direction at time t corresponds to the true direction at time t\u2212\u03c4 (where \u03c4 is the light-time offset). Applying this correction helps account for deviations that can reach several arcseconds.\n\n### Visualize Matched Stars on the Image\n\nTo display matched stars directly on the original astronomical image:\n\n```python\n>>> from starextractor import parse_image\n\n>>> # Load the image (supports FITS or BMP formats)\n>>> image_file = 'obs/fits/img_00000.fits' # or 'obs/bmp/img_00000.bmp'\n>>> image_data = parse_image(image_file)\n\n>>> # Extract sources from the image\n>>> sources = image_data.find_source()\n\n>>> # Configure camera parameters\n>>> camera_params = {'fov': (8, 8), 'pixel_width': 0.01, 'res': (1024, 1024)}\n>>> sources = StarMatch.from_sources(sources.xy, camera_params, flux_raw=sources.brightness)\n\n>>> # Perform star map alignment using a known approximate orientation\n>>> fp_radec = [201, 31] # Example pointing direction [RA, Dec] in degrees\n>>> sources.align(fp_radec, sc_simplified)\n\n>>> # Display matched stars on the image\n>>> sources.show_starmatch(image_data.image_raw, image_data._offset)\n```\n\n<p align=\"middle\">\n <img src=\"readme_figs/output_52_0.png\" width=\"500\" />\n</p>\n\nThe stars marked in the image should align with entries in `catalog_df`:\n\n```python\nprint(sources.matched_results.catalog_df)\n```\n\n<p align=\"middle\">\n <img src=\"readme_figs/catalog_df.png\" width=\"700\" />\n</p>\n\n### Troubleshooting on Star Map Matching\n\nIf star map matching fails, consider the following:\n\n1. **Image Coordinates**: Ensure that the origin of the image coordinates is correctly set. The origin may differ (e.g., upper-left corner vs. lower-left corner). Adjust the pixel coordinates of the sources accordingly.\n\n2. **Field of View**: Select an appropriate star catalog based on the field of view:\n \n - **Bright star catalog**: Suitable for a large field of view.\n \n - **Faint star catalog**: Suitable for a smaller field of view.\n\n3. **Geometric Invariants**: When constructing geometric invariants, the default number of stars used in the **kd-Tree nearest neighbor search** is 9. Increasing this number may **improve the success rate of matching**.\n\n4. **Blind Matching**: For small fields of view, consider increasing the **HEALPix level** to enhance the success rate of blind matching.\n\n### Geometric Distortion Model\n\nAvailable distortion models include Standard Radial Distortion Model(SRDM), Division-mode Radial Distortion Model(DRDM), Tangential Distortion Model(also known as the de-centering distortion), and Brown-Conrady Distortion Model(BCDM).\n\nConsidering that the distortion model involves the power of the distance from the distortion center to pixels, the length scale is introduced here. Associaated with the length scale, the normalized model coefficients is also necessary.\n\n#### Example: Standard Radial Distortion Model(SRDM)\n\nThe Standard Radial Distortion Model(SRDM) is defined by the distortion center and the distortion coefficients.\n\n- The 3rd order polynomial in standard form: $r_u = r_d + k_1 r_d^3$\n- The 5th order polynomial in standard form: $r_u = r_d + k_1 r_d^3 + k_2 r_d^5$\n- The 5th order polynomial in all form: $r_u = r_d + k_1 r_d^2 + k_2 r_d^3 + k_3 r_d^4 + k_4 r_d^5$,\n where $r_d$ is the distance between the distorted pixel coordinates and the distortion center, $r_u$ is the distance between the distortion-corrected pixel coordinates and the distortion center.\n1. The 3rd order polynomial in standard form only works well for small amounts of distortion.\n2. The 5th order polynomial produce more accurate results, especially for \u201cwave\u201d or \u201cmustache\u201d distortion, which might resemble barrel near the center of the image and pincushion near the corners (or vice-versa).\n3. The 5th order all form use all coefficients up to the maximum instead of alternate coefficients (odd or even-only). We have not observed much advantage to these settings.\n4. Higher order polynomials (7th order or higher) should be used with great care because results can become unstable, especially at the outer parts of the image. The image used to calculate the coefficients should have valid corner points near the edge of the image and there should be sufficient rows or columns. \n\nBasic formulas are as follows:\n\n$x_u - x_c = r_u \\cos(\\theta) = r_u (x_d - x_c)/r_d = ... $\n\n$y_u - y_c = r_u \\sin(\\theta) = r_u (y_d - y_c)/r_d = ... $\n\nwhere $(x_c,y_c)$ is the pixel coordinates of the distortion center.\n\nFor more details, please refer to \n\n1. [Distortion_(optics)](https://en.wikipedia.org/wiki/Distortion_(optics))\n2. [imatest distortion-methods](https://www.imatest.com/docs/distortion-methods-and-modules/)\n3. [imatest distortion-models](https://www.imatest.com/support/docs/pre-5-2/geometric-calibration-deprecated/distortion-models/) \n\n##### Construct a Standard Radial Distortion Model\n\n```python\n>>> from starmatch.classes import Distortion\n>>> model = 'RadialStandard' # Type of distortion model\n>>> coeffs = [-1e-4,1e-4] # Coefficients of 5th order SRDM in form of [k1,k2]\n>>> dc = [0.1,0.1] # Pixel coordinates of the distortion center\n>>> distortion_scale = 128 # The length scale of the distortion model, that is, the number of pixels per unit length\n>>> distortion = Distortion(model,coeffs,dc,distortion_scale) # Establish a distortion model\n```\n\n##### Compute the distortion-corrected pixel coordinates\n\n```python\n>>> pixels_xy = [367,125]\n>>> pixels_XY = distortion.apply(pixels_xy)\n>>> print(pixels_XY)\n>>> # [[369.21610677 125.70199655]]\n>>> #Calculate the distortion-corrected pixel coordinates at normalized scale\n>>> #pixels_xy = [[2.8671875, 0.9765625], [1.109375, -0.875]]\n>>> #pixel_scale = 128\n>>> #pixels_XY = distortion.apply(pixels_xy,pixel_scale)\n>>> #print(pixels_XY)\n```\n\n##### Sketch the distortion\n\nVector plot of the distortion:\n\n```python\n>>> xlim,ylim = 512,512\n>>> distortion.sketchmap(xlim,ylim)\n>>> #For normalized pixel coordinates\n>>> #xlim,ylim = 4,4\n>>> #pixel_scale = 128\n>>> #distortion.sketchmap(xlim,ylim,pixel_scale)\n```\n\n<p align=\"middle\">\n <img src=\"readme_figs/output_8_1.png\" width=\"500\" />\n</p>\n\nContour plot of the distortion in x and y components respectively:\n\n```python\n>>> distortion.sketchmap(xlim,ylim,mode='contourf')\n```\n\n<p align=\"middle\">\n <img src=\"readme_figs/output_8_1_contourf.png\" width=\"800\" />\n</p>\n\n#### Example: Division-mode Radial Distortion Model(DRDM)\n\nThe Division-mode Radial Distortion Model defined by the distortion center and the distortion coefficients.\n\n- The 2nd order in division form: $r_u = r_d /(1+ k_1 r_d^2)$\n- The 4th order in division form: $r_u = r_d /(1+ k_1 r_d^2 + k_2 r_d^4)$\n\nBasic formulas for DRDM are same as that for SRDM.\n\n##### Construct a Division-mode Radial Distortion Model\n\n```python\n>>> from starmatch.classes import Distortion\n>>> model = 'RadialDivision' # Type of distortion model\n>>> coeffs = [-1e-4,1e-4] # Coefficients of 4th order DRDM in form of [k1,k2]\n>>> dc = [0.1,0.1] # Pixel coordinates of the distortion center\n>>> distortion_scale = 128 # The length scale of the distortion model, that is, the number of pixels per unit length\n>>> distortion = Distortion(model,coeffs,dc,distortion_scale) # Establish a distortion model\n```\n\n##### Compute the distortion-corrected pixel coordinates\n\n```python\n>>> pixels_xy = [367,125]\n>>> pixels_XY = distortion.apply(pixels_xy)\n>>> print(pixels_XY)\n>>> # [[364.79767243 124.30236829]]\n>>> #Calculate the distortion-corrected pixel coordinates at normalized scale\n>>> #pixels_xy = [[2.8671875, 0.9765625], [1.109375, -0.875]]\n>>> #pixel_scale = 128\n>>> #pixels_XY = distortion.apply(pixels_xy,pixel_scale)\n>>> #print(pixels_XY)\n```\n\n##### Sketch the vector plot of distortion\n\n```python\n>>> xlim,ylim = 512,512\n>>> distortion.sketchmap(xlim,ylim)\n>>> #For normalized pixel coordinates\n>>> #xlim,ylim = 4,4\n>>> #pixel_scale = 128\n>>> #distortion.sketchmap(xlim,ylim,pixel_scale)\n```\n\n<p align=\"middle\">\n <img src=\"readme_figs/output_15_1.png\" width=\"500\" />\n</p>\n\n#### Example: Tangential Distortion Model(TDM)\n\nThe Tangential Distortion Model(also known as the de-centering distortion) is defined by the distortion center and the distortion coefficients.\n\nBasic formulas are as follows:\n\n$x_u = x_d + (P_1 (r_d^2 + 2 (x_d - x_c)^2) + 2 P_2 (x_d-x_c) (y_d-y_c)) (1 + P_3 r_d^2 + P_4 r_d^4 + ...)$\n\n$y_u = y_d + (P_2 (r_d^2 + 2 (y_d - y_c)^2) + 2 P_1 (x_d-x_c) (y_d-y_c)) (1 + P_3 r_d^2 + P_4 r_d^4 + ...)$\n\n##### Construct a Tangential Distortion Model\n\n```python\n>>> from starmatch.classes import Distortion\n>>> model = 'Tangential' # Type of distortion model\n>>> coeffs = [-1e-4,1e-4] # Coefficients of 2nd order SRDM in form of [P1,P2]\n>>> dc = [0.1,0.1] # Pixel coordinates of the distortion center\n>>> distortion_scale = 128 # The length scale of the distortion model, that is, the number of pixels per unit length\n>>> distortion = Distortion(model,coeffs,dc,distortion_scale) # Establish a distortion model\n```\n\n##### Compute the distortion-corrected pixel coordinates\n\n```python\n>>> pixels_xy = [367,125]\n>>> pixels_XY = distortion.apply(pixels_xy)\n>>> print(pixels_XY)\n>>> # [[366.75821931 125.06542319]]\n>>> #Calculate the distortion-corrected pixel coordinates at normalized scale\n>>> #pixels_xy = [[2.8671875, 0.9765625], [1.109375, -0.875]]\n>>> #pixel_scale = 128\n>>> #pixels_XY = distortion.apply(pixels_xy,pixel_scale)\n>>> #print(pixels_XY)\n```\n\n##### Sketch the vector plot of distortion\n\n```python\n>>> xlim,ylim = 512,512\n>>> distortion.sketchmap(xlim,ylim)\n>>> #For normalized pixel coordinates\n>>> #xlim,ylim = 4,4\n>>> #pixel_scale = 128\n>>> #distortion.sketchmap(xlim,ylim,pixel_scale)\n```\n\n<p align=\"middle\">\n <img src=\"readme_figs/output_22_1.png\" width=\"500\" />\n</p>\n\n#### Example: Brown\u2013Conrady Distortion Model(BCDM)\n\nThe Brown\u2013Conrady model corrects both the radial distortion and the tangential distortion caused by physical elements in a lens not being perfectly aligned.\n\nBasic formulas are as follows:\n\n$x_u = x_d + (x_d - x_c) (K_1 r_d^2 + K_2 r_d^4 + ...) + (P_1 (r_d^2 + 2 (x_d-x_c)^2) + 2 P_2 (x_d-x_c) (y_d-y_c)) (1 + P_3 r_d^2 + P_4 r_d^4 + ...)$\n\n$y_u = y_d + (y_d - x_c) (K_1 r_d^2 + K_2 r_d^4 + ...) + (P_2 (r_d^2 + 2 (y_d-y_c)^2) + 2 P_1 (x_d-x_c) (y_d-y_c)) (1 + P_3 r_d^2 + P_4 r_d^4 + ...)$\n\n##### Construct a Brown\u2013Conrady Distortion Model\n\n```python\n>>> from starmatch.classes import Distortion\n>>> model = 'Brown\u2013Conrady' # Type of distortion model\n>>> coeffs = [[-1e-4,1e-4],[1e-3,1e-3,1e-4,1e-5]] # Coefficients of Brown\u2013Conrady distortion model in form of [[coeffs_radial],[coeffs_tangential]]\n>>> dc = [0.1,0.1] # Pixel coordinates of the distortion center\n>>> distortion_scale = 128 # The length scale of the distortion model, that is, the number of pixels per unit length\n>>> distortion = Distortion(model,coeffs,dc,distortion_scale) # Establish a distortion model\n```\n\n### Compute the distortion-corrected pixel coordinates\n\n```python\n>>> pixels_xy = [367,125]\n>>> pixels_XY = distortion.apply(pixels_xy)\n>>> print(pixels_XY)\n>>> # [[372.88150908 127.60108593]]\n>>> #Calculate the distortion-corrected pixel coordinates at normalized scale\n>>> #pixels_xy = [[2.8671875, 0.9765625], [1.109375, -0.875]]\n>>> #pixel_scale = 128\n>>> #pixels_XY = distortion.apply(pixels_xy,pixel_scale)\n>>> #print(pixels_XY)\n```\n\n##### Sketch the vector plot of distortion\n\n```python\n>>> xlim,ylim = 512,512\n>>> distortion.sketchmap(xlim,ylim)\n>>> #For normalized pixel coordinates\n>>> #xlim,ylim = 4,4\n>>> #pixel_scale = 128\n>>> #distortion.sketchmap(xlim,ylim,pixel_scale)\n```\n\n<p align=\"middle\">\n <img src=\"readme_figs/output_29_1.png\" width=\"500\" />\n</p>\n\n## Change log\n\n- **0.2.7 \u2014 Nov 19, 2024**\n\n - Previously, the star map matching process was based on two iterations. An additional matching iteration has been introduced, significantly improving the accuracy of star map matching.\n\n- **0.2.6 \u2014 Nov 18, 2024**\n \n - Implemented different pixel tolerances for initial and secondary matching stages, improving the success rate for both star map matching and blind matching.\n\n- **0.2.5 \u2014 Oct 30, 2024**\n \n - Improved outliers identification in star map matching with the method of LOWESS (Locally Weighted Scatterplot Smoothing).\n\n- **0.2.4 \u2014 Sep 29, 2024**\n \n - In order to fit the distortion model, it is necessary to select as many stars as possible in the field of view. However, considering the computational speed budget, the selection of all stars is modified to select evenly distributed bright stars.\n - The RLM(Robust Linear Model) is used instead of ordinary least squares to estimate the magnitude constant C and its uncertainty\n based on observed fluxes and apparent magnitudes.\n\n- **0.2.3 \u2014 Sep 03, 2024**\n \n - Improved WCS transform.\n - Optimized threshold parameters.\n\n- **0.2.2 \u2014 Aug 27, 2024**\n \n - By adding a secondary test, the error rate of blind star map matching is significantly reduced.\n\n- **0.2.1 \u2014 Jul 09, 2024**\n \n - Added two methods for outliers recognition in star map mismatching ahead of distortion calibration:\n - 'lowess': Identifies outliers with the method of LOWESS (Locally Weighted Scatterplot Smoothing). LOWESS uses a weighted **linear regression** by default.\n - 'iqr': Identifies outliers with the method of Interquartile Range (IQR).\n\n- **0.2.0 \u2014 Jul 07, 2024**\n \n - Compatible with astronomical corrections, including proper motion, annual parallax, aberration, and light deflection.\n - Added two distortion estimation methods:\n - 'piecewise-affine': The transform is based on a Delaunay triangulation of the points to form a mesh. Each triangle is used to find a local affine transform.\n - 'polynomial': 2D polynomial transformation.\n - Implemented a star map blind matching algorithm for astronomical images, which can adapt to the field of view of images from tens of degrees to tens of arc minutes.\n\n- **0.1.4 \u2014 Sep 23, 2023**\n \n - Fixed an issue where star chart matching failed after performing a blind match.\n\n- **0.1.3 \u2014 Sep 05, 2023**\n \n - Added contour plot for distortion models.\n - Minor bugs fixed.\n\n- **0.1.2 \u2014 Jul 23, 2023**\n \n - Simplified the use of blind matching.\n - Added pixel width and field of view estimates for blind matching.\n\n- **0.1.1 \u2014 Jun 16, 2023**\n \n - The ***starmatch*** package was released.\n\n## Next release\n\n- Find the inverse transformation of distortion models\n\n## Reference\n\n- [Distortion_(optics)](https://en.wikipedia.org/wiki/Distortion_(optics))\n- [imatest distortion-methods](https://www.imatest.com/docs/distortion-methods-and-modules/)\n- [imatest distortion-models](https://www.imatest.com/support/docs/pre-5-2/geometric-calibration-deprecated/distortion-models/) \n- [Astroalign](https://astroalign.quatrope.org/en/latest/index.html)\n",
"bugtrack_url": null,
"license": "MIT",
"summary": "A package to handle the Star chart matching and astrometric positioning",
"version": "0.2.7",
"project_urls": {
"Homepage": "https://github.com/lcx366/STARMATCH"
},
"split_keywords": [
"star map matching",
" astrometric positioning",
" distortion calibration",
" astronomical correction"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "819385129ee4c805e15b7547d56f2b78da0e55f0a5539a4a7a26398e8d345286",
"md5": "8f44da3ca80062f9ea049fadb6f5b081",
"sha256": "68dcfde795aa586cfebd0807692d046485e0b5566befdd1c48680c1a045da18d"
},
"downloads": -1,
"filename": "starmatch-0.2.7-py3-none-any.whl",
"has_sig": false,
"md5_digest": "8f44da3ca80062f9ea049fadb6f5b081",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.10",
"size": 34612,
"upload_time": "2024-11-19T11:47:53",
"upload_time_iso_8601": "2024-11-19T11:47:53.292671Z",
"url": "https://files.pythonhosted.org/packages/81/93/85129ee4c805e15b7547d56f2b78da0e55f0a5539a4a7a26398e8d345286/starmatch-0.2.7-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-11-19 11:47:53",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "lcx366",
"github_project": "STARMATCH",
"travis_ci": false,
"coveralls": false,
"github_actions": false,
"lcname": "starmatch"
}
JSON
