### Command-Line Examples

Source: https://github.com/steffenhir/graxpert/blob/main/README.md

Illustrative examples of how to use GraXpert from the command line.

```APIDOC
## Command-Line Examples

These examples demonstrate GraXpert's command-line usage. Note the executable name may vary based on your OS (e.g., `GraXpert-linux`, `GraXpert.app/Contents/MacOS/GraXpert`).

### Basic Usage

Performs default background extraction on `my_image.fits`.
```bash
GraXpert-win64.exe my_image.fits -cli
```

### Advanced Background Extraction

Applies background extraction using AI version '1.1', 'Division' correction, '0.1' smoothing, and saves the background model.
```bash
GraXpert-win64.exe my_image.fits -cli -ai_version 1.1 -correction Division -smoothing 0.1 -bg
```

### Denoising Example

Applies denoising with a strength of 0.8 and a batch size of 8.
```bash
GraXpert-win64.exe my_image.fits -cli -cmd denoising -strength 0.8 -batch_size 8
```
```

--------------------------------

### Install Requirements (Windows/Linux)

Source: https://github.com/steffenhir/graxpert/blob/main/README.md

Install all project dependencies listed in 'requirements.txt' using pip on Windows or Linux. Ensure the virtual environment is activated.

```bash
pip install -r requirements.txt
```

--------------------------------

### Install Requirements (macOS)

Source: https://github.com/steffenhir/graxpert/blob/main/README.md

Install all project dependencies listed in 'requirements.txt' using pip3 on macOS. Ensure the virtual environment is activated.

```bash
pip3 install -r requirements.txt
```

--------------------------------

### Run GraXpert Application

Source: https://github.com/steffenhir/graxpert/blob/main/README.md

Execute the GraXpert main module using Python. This command should be run after setting up the virtual environment and installing all required packages.

```bash
python -m graxpert.main
```

--------------------------------

### Check GraXpert Version CLI

Source: https://context7.com/steffenhir/graxpert/llms.txt

Display the installed version of the GraXpert CLI tool.

```bash
GraXpert-linux -v
```

--------------------------------

### Create Python Virtual Environment (Linux/macOS)

Source: https://github.com/steffenhir/graxpert/blob/main/README.md

Use this command to create a new virtual environment named 'graxpert-env' on Linux or macOS. Ensure Python 3.10 or later is installed.

```bash
python3 -m venv graxpert-env
```

--------------------------------

### Create Python Virtual Environment (Windows)

Source: https://github.com/steffenhir/graxpert/blob/main/README.md

Use this command to create a new virtual environment named 'graxpert-env' on Windows. Ensure Python 3.10 or later is installed.

```bash
python -m venv graxpert-env
```

--------------------------------

### Install Tkinter (macOS)

Source: https://github.com/steffenhir/graxpert/blob/main/README.md

For macOS users, install the Python Tkinter library using Homebrew, specifically version 3.10, to resolve potential compatibility issues with newer macOS versions like Sonoma.

```bash
brew install python-tk@3.10
```

--------------------------------

### Clone the GraXpert repository

Source: https://github.com/steffenhir/graxpert/blob/main/README.md

Download the source code from GitHub to begin development.

```bash
git clone https://github.com/Steffenhir/GraXpert
cd GraXpert
```

--------------------------------

### Manage Preferences

Source: https://context7.com/steffenhir/graxpert/llms.txt

Imports for handling application preferences and configuration files.

```python
from graxpert.preferences import Prefs, load_preferences, save_preferences
import os
from appdirs import user_config_dir
```

--------------------------------

### Make GraXpert executable on Linux

Source: https://github.com/steffenhir/graxpert/blob/main/README.md

Set the execution permission for the GraXpert binary on Linux systems.

```bash
chmod u+x ./GraXpert-linux
```

--------------------------------

### Complete Image Processing Pipeline

Source: https://context7.com/steffenhir/graxpert/llms.txt

Demonstrates a full image processing workflow including loading, background extraction, denoising using AI models, and saving. Ensure necessary GraXpert modules and numpy are imported.

```python
from graxpert.astroimage import AstroImage
from graxpert.background_extraction import extract_background
from graxpert.denoising import denoise
from graxpert.ai_model_handling import (
    ai_model_path_from_version,
    bge_ai_models_dir,
    denoise_ai_models_dir,
    latest_version,
    download_version,
    list_local_versions
)
from graxpert.s3_secrets import bge_bucket_name, denoise_bucket_name
import numpy as np

# Load the astronomical image
input_image = AstroImage(do_update_display=False)
input_image.set_from_file("/path/to/raw_image.fits", None, None)

# Ensure AI models are available
bge_version = latest_version(bge_ai_models_dir, bge_bucket_name)
if bge_version not in [v["version"] for v in list_local_versions(bge_ai_models_dir)]:
    download_version(bge_ai_models_dir, bge_bucket_name, bge_version)

denoise_version = latest_version(denoise_ai_models_dir, denoise_bucket_name)
if denoise_version not in [v["version"] for v in list_local_versions(denoise_ai_models_dir)]:
    download_version(denoise_ai_models_dir, denoise_bucket_name, denoise_version)

# Step 1: Background Extraction
bge_model_path = ai_model_path_from_version(bge_ai_models_dir, bge_version)
background = extract_background(
    in_imarray=input_image.img_array,
    background_points=np.array([]),
    interpolation_type="AI",
    smoothing=0.1,
    downscale_factor=1,
    sample_size=25,
    RBF_kernel="thin_plate",
    spline_order=3,
    corr_type="Subtraction",
    ai_path=bge_model_path,
    ai_gpu_acceleration=True
)

# Step 2: Denoising
denoise_model_path = ai_model_path_from_version(denoise_ai_models_dir, denoise_version)
denois ed_array = denoise(
    image=input_image.img_array,
    ai_path=denoise_model_path,
    strength=0.5,
    batch_size=4,
    ai_gpu_acceleration=True
)
```

--------------------------------

### Activate Virtual Environment (Windows)

Source: https://github.com/steffenhir/graxpert/blob/main/README.md

Activate the 'graxpert-env' virtual environment on Windows to isolate project dependencies. This command should be run after creating the environment.

```bash
graxpert-env\Scripts\activate
```

--------------------------------

### Activate Virtual Environment (Linux/macOS)

Source: https://github.com/steffenhir/graxpert/blob/main/README.md

Activate the 'graxpert-env' virtual environment on Linux or macOS to isolate project dependencies. This command should be run after creating the environment.

```bash
source graxpert-env/bin/activate
```

--------------------------------

### Execute basic background extraction via CLI

Source: https://github.com/steffenhir/graxpert/blob/main/README.md

Run the default background extraction on an image file using the command-line interface.

```bash
GraXpert-win64.exe my_image.fits -cli
```

--------------------------------

### Load and Modify User Preferences

Source: https://context7.com/steffenhir/graxpert/llms.txt

Demonstrates loading preferences from a JSON file, viewing current settings, modifying them, and saving the changes. This is useful for interactive adjustments.

```python
prefs_filename = os.path.join(user_config_dir(appname="GraXpert"), "preferences.json")
prefs = load_preferences(prefs_filename)

# View current settings
print(f"Interpolation type: {prefs.interpol_type_option}")
print(f"Smoothing: {prefs.smoothing_option}")
print(f"Correction type: {prefs.corr_type}")
print(f"GPU acceleration: {prefs.ai_gpu_acceleration}")

# Modify preferences for processing
prefs.interpol_type_option = "AI"
prefs.smoothing_option = 0.2
prefs.corr_type = "Subtraction"
prefs.bge_ai_version = "1.1.0"
prefs.denoise_strength = 0.6
prefs.deconvolution_strength = 0.5
prefs.deconvolution_psfsize = 3.5
prefs.ai_batch_size = 8
prefs.ai_gpu_acceleration = True
prefs.saveas_option = "32 bit Fits"

# Save preferences
save_preferences(prefs_filename, prefs)
```

--------------------------------

### GraXpert Command-Line Arguments

Source: https://github.com/steffenhir/graxpert/blob/main/README.md

Overview of the general command-line arguments available for GraXpert's AI operations.

```APIDOC
## GraXpert Command-Line Usage

GraXpert offers a command-line interface for its AI-powered operations, allowing for automated processing without the graphical user interface.

### General Arguments

- **-cmd [image_operation]** (string) - Specifies the AI operation to perform. Options: "background-extraction" (default), "denoising".
- **filename** (string) - The path to the input image file (required).
- **-cli** (flag) - Must be included to enable command-line integration; otherwise, the GUI will launch.
- **-output [output_file_name]** (string) - Specifies the name for the output image file (without extension). If omitted, '_GraXpert' is appended to the original filename.
- **-preferences_file** (string) - Path to a preferences file containing background grid points for extraction methods.
- **-gpu** (boolean) - Set to 'false' to disable GPU acceleration for AI inference. Defaults to 'true' to enable it.
- **-ai_version [version]** (string) - Specifies the AI model version to use. Defaults to the latest available version. Can be a locally or remotely available version.
```

--------------------------------

### Apply Image Stretching with Presets

Source: https://context7.com/steffenhir/graxpert/llms.txt

Applies Midtone Transfer Function (MTF) stretching to astronomical images using predefined stretch options. Ensure numpy and graxpert.stretch are imported.

```python
from graxpert.stretch import stretch, StretchParameters, MTF
import numpy as np

# Create stretch parameters with preset
stretch_params = StretchParameters(
    stretch_option="15% Bg, 3 sigma",  # "No Stretch", "10% Bg, 3 sigma",
                                        # "15% Bg, 3 sigma", "20% Bg, 3 sigma",
                                        # "30% Bg, 2 sigma"
    channels_linked=False,              # Process channels independently
    images_linked=False
)

# Apply stretch to image
linear_image = np.random.rand(1000, 1500, 3).astype(np.float32) * 0.3  # Dark image
stretched_image = stretch(linear_image, stretch_params)

# Manual MTF calculation for custom stretching
data = np.random.rand(100, 100).astype(np.float32)
midtone = 0.25  # Target midtone level
stretched = MTF(data.copy(), midtone)

# Access stretch parameters
print(f"Background level: {stretch_params.bg}")
print(f"Sigma: {stretch_params.sigma}")
print(f"Do stretch: {stretch_params.do_stretch}")
```

--------------------------------

### GraXpert CLI Commands

Source: https://context7.com/steffenhir/graxpert/llms.txt

This section details the various commands available through the GraXpert command-line interface for batch processing and automation.

```APIDOC
## GraXpert CLI Commands

### Description
Provides a comprehensive CLI for automated image processing without the GUI. Supports background extraction, denoising, and deconvolution operations with configurable AI model versions and processing parameters.

### Basic Background Extraction
```bash
GraXpert-linux my_image.fits -cli
```

### Advanced Background Extraction
```bash
GraXpert-linux my_image.fits -cli -cmd background-extraction \
    -ai_version 1.1.0 \
    -correction Division \
    -smoothing 0.1 \
    -bg \
    -output processed_image
```

### Denoising
```bash
GraXpert-linux my_image.fits -cli -cmd denoising \
    -strength 0.7 \
    -batch_size 4 \
    -gpu true
```

### Object Deconvolution
```bash
GraXpert-linux my_image.fits -cli -cmd deconv-obj \
    -strength 0.5 \
    -psfsize 3.0 \
    -batch_size 4
```

### Stellar Deconvolution
```bash
GraXpert-linux my_image.fits -cli -cmd deconv-stellar \
    -strength 0.6 \
    -psfsize 2.5 \
    -output sharpened_stars
```

### Check Version
```bash
GraXpert-linux -v
```
```

--------------------------------

### Stellar Deconvolution CLI

Source: https://context7.com/steffenhir/graxpert/llms.txt

Sharpen stars using stellar deconvolution via the CLI, specifying strength, PSF size, and output file.

```bash
GraXpert-linux my_image.fits -cli -cmd deconv-stellar \
    -strength 0.6 \
    -psfsize 2.5 \
    -output sharpened_stars
```

--------------------------------

### Basic AI Background Extraction CLI

Source: https://context7.com/steffenhir/graxpert/llms.txt

Perform basic background extraction using the AI model via the command-line interface. No manual sample points are required.

```bash
GraXpert-linux my_image.fits -cli
```

--------------------------------

### Manage AI Models

Source: https://context7.com/steffenhir/graxpert/llms.txt

Provides utilities to list, download, and validate AI model versions from local or remote storage.

```python
from graxpert.ai_model_handling import (
    list_local_versions,
    list_remote_versions,
    download_version,
    latest_version,
    ai_model_path_from_version,
    bge_ai_models_dir,
    denoise_ai_models_dir,
    deconvolution_object_ai_models_dir,
    deconvolution_stars_ai_models_dir
)
from graxpert.s3_secrets import (
    bge_bucket_name,
    denoise_bucket_name,
    deconvolution_object_bucket_name,
    deconvolution_stars_bucket_name
)

# List available local AI model versions
local_bge_models = list_local_versions(bge_ai_models_dir)
print(f"Local BGE models: {[m['version'] for m in local_bge_models]}")

local_denoise_models = list_local_versions(denoise_ai_models_dir)
print(f"Local denoise models: {[m['version'] for m in local_denoise_models]}")

# List available remote versions
remote_bge_models = list_remote_versions(bge_bucket_name)
print(f"Remote BGE models: {[m['version'] for m in remote_bge_models]}")

# Get latest version (checks both local and remote)
latest_bge = latest_version(bge_ai_models_dir, bge_bucket_name)
print(f"Latest BGE version: {latest_bge}")

# Download a specific version
download_version(
    ai_models_dir=denoise_ai_models_dir,
    bucket_name=denoise_bucket_name,
    target_version="1.1.0"
)

# Get model path for a version
model_path = ai_model_path_from_version(bge_ai_models_dir, "1.1.0")
# Returns: /path/to/bge-ai-models/1.1.0/model.onnx
```

--------------------------------

### Denoising with Strength Control CLI

Source: https://context7.com/steffenhir/graxpert/llms.txt

Apply AI-powered denoising to an image using the CLI, controlling the denoising strength and batch size for GPU processing.

```bash
GraXpert-linux my_image.fits -cli -cmd denoising \
    -strength 0.7 \
    -batch_size 4 \
    -gpu true
```

--------------------------------

### Customized Background Extraction CLI

Source: https://context7.com/steffenhir/graxpert/llms.txt

Execute background extraction with custom AI model version, correction type, and smoothing parameters. Specify output file.

```bash
GraXpert-linux my_image.fits -cli -cmd background-extraction \
    -ai_version 1.1.0 \
    -correction Division \
    -smoothing 0.1 \
    -bg \
    -output processed_image
```

--------------------------------

### Background Extraction Command-Line Options

Source: https://github.com/steffenhir/graxpert/blob/main/README.md

Specific arguments for the background extraction AI operation.

```APIDOC
### Background Extraction Specific Arguments

- **-correction [type]** (string) - Selects the background correction method. Options: "Subtraction" (default), "Division".
- **-smoothing [strength]** (float) - Adjusts the strength of smoothing applied during background extraction. Range: 0.0 (no smoothing) to 1.0 (maximum smoothing).
- **-bg** (flag) - If present, the generated background model will also be saved.
```

--------------------------------

### Execute advanced background extraction via CLI

Source: https://github.com/steffenhir/graxpert/blob/main/README.md

Perform background extraction with specific AI version, correction type, smoothing, and background model saving.

```bash
GraXpert-win64.exe my_image.fits -cli -ai_version 1.1 -correction Division -smoothing 0.1 -bg
```

--------------------------------

### Denoising Command-Line Options

Source: https://github.com/steffenhir/graxpert/blob/main/README.md

Specific arguments for the denoising AI operation.

```APIDOC
### Denoising Specific Arguments

- **-strength [value]** (float) - Adjusts the strength of the denoising effect. Range: 0.0 (minimum) to 1.0 (maximum). Default: "0.5".
- **-batch_size [value]** (integer) - Sets the number of image tiles to denoise in parallel. Warning: Increasing this value may lead to out-of-memory errors. Valid Range: 1 to 32. Default: "4".
```

--------------------------------

### Create Preferences JSON for CLI Batch Processing

Source: https://context7.com/steffenhir/graxpert/llms.txt

Generates a JSON file containing preferences specifically for command-line interface (CLI) batch processing. This allows for reproducible and automated workflows.

```python
import json

batch_prefs = {
    "interpol_type_option": "RBF",
    "RBF_kernel": "thin_plate",
    "smoothing_option": 0.3,
    "corr_type": "Subtraction",
    "sample_size": 25,
    "background_points": [[100, 200], [500, 300], [900, 400]],
    "ai_gpu_acceleration": True
}

with open("batch_preferences.json", "w") as f:
    json.dump(batch_prefs, f)
```

--------------------------------

### Object Deconvolution CLI

Source: https://context7.com/steffenhir/graxpert/llms.txt

Perform object deconvolution for nebulae and galaxies using the CLI, adjusting strength and PSF size.

```bash
GraXpert-linux my_image.fits -cli -cmd deconv-obj \
    -strength 0.5 \
    -psfsize 3.0 \
    -batch_size 4
```

--------------------------------

### Manage Astronomical Images

Source: https://context7.com/steffenhir/graxpert/llms.txt

Handles loading, processing, and saving of astronomical images including FITS header manipulation.

```python
from graxpert.astroimage import AstroImage
from graxpert.stretch import StretchParameters

# Load an astronomical image
astro_img = AstroImage(do_update_display=False)
astro_img.set_from_file(
    directory="/path/to/image.fits",
    stretch_params=None,
    saturation=1.0
)

# Access image properties
print(f"Image dimensions: {astro_img.width} x {astro_img.height}")
print(f"Number of channels: {astro_img.img_array.shape[-1]}")
print(f"Image format: {astro_img.img_format}")

# Access and modify FITS header
if astro_img.fits_header:
    print(f"Object: {astro_img.fits_header.get('OBJECT', 'Unknown')}")
    astro_img.fits_header['COMMENT'] = 'Processed with GraXpert'

# Create output image from processed array
output_img = AstroImage(do_update_display=False)
output_img.set_from_array(processed_array)
output_img.fits_header = astro_img.fits_header

# Save in different formats
output_img.save("/path/to/output.fits", "32 bit Fits")
output_img.save("/path/to/output.xisf", "32 bit XISF")
output_img.save("/path/to/output.tiff", "16 bit Tiff")

# Crop image
astro_img.crop(startx=100, endx=900, starty=50, endy=750)

# Get local median at a point (useful for sample point selection)
median_rgb = astro_img.get_local_median([500, 300])
```

--------------------------------

### Denoising Python

Source: https://context7.com/steffenhir/graxpert/llms.txt

Apply AI-powered denoising to an image array. Specify the AI model path, denoising strength, and batch size for GPU acceleration.

```python
from graxpert.denoising import denoise
import numpy as np

# Load image data (normalized 0-1 float32)
image_array = np.random.rand(2000, 3000, 3).astype(np.float32)
ai_model_path = "/path/to/denoise-ai-models/1.0.0/model.onnx"

# Apply denoising with medium strength
denois ed_image = denoise(
    image=image_array,
    ai_path=ai_model_path,
    strength=0.5,              # 0.0 (minimum) to 1.0 (maximum)
    batch_size=4,              # Tiles processed in parallel (1-32, power of 2)
    window_size=256,           # Tile size
    stride=128,                # Tile overlap
    ai_gpu_acceleration=True
)
```

--------------------------------

### Perform AI Deconvolution

Source: https://context7.com/steffenhir/graxpert/llms.txt

Applies AI-based deconvolution for sharpening nebulae, galaxies, or stars. Requires specifying the model path and PSF size.

```python
from graxpert.deconvolution import deconvolve
import numpy as np

# Load image data
image_array = np.random.rand(2000, 3000, 3).astype(np.float32)

# Object deconvolution for nebulae/galaxies
obj_model_path = "/path/to/deconvolution-object-ai-models/1.0.1/model.onnx"
deconvolved = deconvolve(
    image=image_array,
    ai_path=obj_model_path,
    strength=0.5,              # 0.0 to 1.0 effect strength
    psfsize=5.0,               # Point spread function size (seeing in pixels)
    batch_size=4,
    window_size=512,
    stride=448,
    ai_gpu_acceleration=True
)

# Stellar deconvolution for sharper stars
stellar_model_path = "/path/to/deconvolution-stars-ai-models/1.0.0/model.onnx"
deconvolved = deconvolve(
    image=image_array,
    ai_path=stellar_model_path,
    strength=0.7,
    psfsize=3.0,               # Typical seeing conditions
    batch_size=4,
    ai_gpu_acceleration=True
)
```

--------------------------------

### Save processed images and background models

Source: https://context7.com/steffenhir/graxpert/llms.txt

Use the AstroImage class to save processed arrays to FITS files. Ensure the FITS header is updated to maintain metadata continuity.

```python
output_image = AstroImage(do_update_display=False)
output_image.set_from_array(denoised_array)
output_image.fits_header = input_image.fits_header
output_image.fits_header['HISTORY'] = 'Background extracted with GraXpert AI'
output_image.fits_header['HISTORY'] = 'Denoised with GraXpert AI'
output_image.fits_header['BGE-AI-VER'] = bge_version
output_image.fits_header['DENOISE-VER'] = denoise_version

output_image.save("/path/to/processed_image.fits", "32 bit Fits")

# Also save the background model
background_image = AstroImage(do_update_display=False)
background_image.set_from_array(background)
background_image.fits_header = input_image.fits_header
background_image.save("/path/to/background_model.fits", "32 bit Fits")

print("Processing complete!")
```

--------------------------------

### Strong Denoising Python

Source: https://context7.com/steffenhir/graxpert/llms.txt

Apply strong AI-powered denoising for very noisy images. Increase batch size for potentially faster GPU processing.

```python
# Strong denoising for very noisy images
denois ed_image = denoise(
    image=image_array,
    ai_path=ai_model_path,
    strength=0.9,
    batch_size=8,              # Increase for faster GPU processing
    ai_gpu_acceleration=True
)
```

--------------------------------

### Perform AI Denoising

Source: https://context7.com/steffenhir/graxpert/llms.txt

Executes AI-based denoising on an image array. Set ai_gpu_acceleration to False for CPU-only processing.

```python
denoised_image = denoise(
    image=image_array,
    ai_path=ai_model_path,
    strength=0.5,
    batch_size=2,
    ai_gpu_acceleration=False
)
```

--------------------------------

### Denoising Function (Python API)

Source: https://context7.com/steffenhir/graxpert/llms.txt

The `denoise` function applies AI-powered noise reduction to astronomical images with adjustable strength.

```APIDOC
## Denoising Function (Python API)

### Description
The `denoise` function applies AI-powered noise reduction to astronomical images with adjustable strength. It processes images in tiles for memory efficiency and supports batch processing for GPU optimization.

### Method
```python
from graxpert.denoising import denoise
import numpy as np
```

### Parameters
- **image** (numpy.ndarray) - Input image data (normalized 0-1 float32).
- **ai_path** (str) - Path to the denoising AI model.
- **strength** (float) - Denoising strength (0.0 to 1.0).
- **batch_size** (int) - Number of tiles processed in parallel (1-32, power of 2).
- **window_size** (int) - Size of the processing tiles.
- **stride** (int) - Overlap between tiles.
- **ai_gpu_acceleration** (bool) - Enable GPU acceleration for AI.

### Request Example (Medium Strength)
```python
image_array = np.random.rand(2000, 3000, 3).astype(np.float32)
ai_model_path = "/path/to/denoise-ai-models/1.0.0/model.onnx"

denois ed_image = denoise(
    image=image_array,
    ai_path=ai_model_path,
    strength=0.5,
    batch_size=4,
    window_size=256,
    stride=128,
    ai_gpu_acceleration=True
)
```

### Request Example (Strong Strength)
```python
image_array = np.random.rand(2000, 3000, 3).astype(np.float32)
ai_model_path = "/path/to/denoise-ai-models/1.0.0/model.onnx"

denois ed_image = denoise(
    image=image_array,
    ai_path=ai_model_path,
    strength=0.9,
    batch_size=8,
    ai_gpu_acceleration=True
)
```
```

--------------------------------

### Background Extraction Function (Python API)

Source: https://context7.com/steffenhir/graxpert/llms.txt

The `extract_background` function is the core processing function that removes gradients from astronomical images using either AI or traditional interpolation methods.

```APIDOC
## Background Extraction Function (Python API)

### Description
The `extract_background` function removes gradients from astronomical images using either AI or traditional interpolation methods. It supports multiple interpolation types and correction methods with configurable smoothing.

### Method
```python
from graxpert.background_extraction import extract_background
import numpy as np
```

### AI-based Background Extraction

#### Parameters
- **in_imarray** (numpy.ndarray) - Input image data.
- **background_points** (numpy.ndarray) - Empty array for AI method.
- **interpolation_type** (str) - Set to "AI".
- **smoothing** (float) - Smoothing factor (0.0 to 1.0).
- **downscale_factor** (int) - Factor to downscale the image for processing.
- **sample_size** (int) - Size of samples for interpolation.
- **RBF_kernel** (str) - Kernel type for RBF interpolation (not used for AI).
- **spline_order** (int) - Order of splines (not used for AI).
- **corr_type** (str) - Correction type: "Subtraction" or "Division".
- **ai_path** (str) - Path to the AI model.
- **ai_gpu_acceleration** (bool) - Enable GPU acceleration for AI.

#### Request Example
```python
image_array = np.random.rand(2000, 3000, 3).astype(np.float32)
ai_model_path = "/path/to/ai-models/1.1.0/model.onnx"

background = extract_background(
    in_imarray=image_array,
    background_points=np.array([]),  # Empty for AI method
    interpolation_type="AI",
    smoothing=0.1,
    downscale_factor=1,
    sample_size=25,
    RBF_kernel="thin_plate",
    spline_order=3,
    corr_type="Subtraction",
    ai_path=ai_model_path,
    ai_gpu_acceleration=True
)
```

### Traditional RBF Interpolation

#### Parameters
- **in_imarray** (numpy.ndarray) - Input image data.
- **background_points** (numpy.ndarray) - Array of [x, y] coordinates for background sample points.
- **interpolation_type** (str) - Set to "RBF", "Splines", or "Kriging".
- **smoothing** (float) - Smoothing factor.
- **downscale_factor** (int) - Factor to downscale the image for processing.
- **sample_size** (int) - Size of samples for interpolation.
- **RBF_kernel** (str) - Kernel type for RBF interpolation: "thin_plate", "linear", "cubic".
- **spline_order** (int) - Order of splines.
- **corr_type** (str) - Correction type: "Subtraction" or "Division".
- **ai_path** (str) - Path to the AI model (set to None for traditional methods).
- **ai_gpu_acceleration** (bool) - Enable GPU acceleration for AI (set to False for traditional methods).

#### Request Example
```python
image_array = np.random.rand(2000, 3000, 3).astype(np.float32)
background_points = np.array([
    [100, 200],
    [500, 300],
    [800, 150],
    [1200, 400],
    [300, 600],
    [900, 550]
])

background = extract_background(
    in_imarray=image_array,
    background_points=background_points,
    interpolation_type="RBF",
    smoothing=0.5,
    downscale_factor=4,
    sample_size=25,
    RBF_kernel="thin_plate",
    spline_order=3,
    corr_type="Subtraction",
    ai_path=None,
    ai_gpu_acceleration=False
)
```
```

--------------------------------

### AI Background Extraction Python

Source: https://context7.com/steffenhir/graxpert/llms.txt

Extract background gradients from an image array using AI. No background sample points are needed. Ensure the AI model path is correct.

```python
from graxpert.background_extraction import extract_background
import numpy as np

# AI-based background extraction (no sample points needed)
image_array = np.random.rand(2000, 3000, 3).astype(np.float32)  # RGB image
ai_model_path = "/path/to/ai-models/1.1.0/model.onnx"

background = extract_background(
    in_imarray=image_array,
    background_points=np.array([]),  # Empty for AI method
    interpolation_type="AI",
    smoothing=0.1,                    # 0.0 to 1.0
    downscale_factor=1,
    sample_size=25,
    RBF_kernel="thin_plate",
    spline_order=3,
    corr_type="Subtraction",          # "Subtraction" or "Division"
    ai_path=ai_model_path,
    ai_gpu_acceleration=True
)
# image_array is modified in-place with gradient removed
```

--------------------------------

### RBF Background Extraction Python

Source: https://context7.com/steffenhir/graxpert/llms.txt

Perform background extraction using Radial Basis Functions (RBF) interpolation with manually selected background sample points. Specify RBF kernel and smoothing.

```python
# Traditional RBF interpolation with manual sample points
background_points = np.array([
    [100, 200],
    [500, 300],
    [800, 150],
    [1200, 400],
    [300, 600],
    [900, 550]
])

background = extract_background(
    in_imarray=image_array,
    background_points=background_points,
    interpolation_type="RBF",         # "RBF", "Splines", or "Kriging"
    smoothing=0.5,
    downscale_factor=4,               # Downscale for faster processing
    sample_size=25,
    RBF_kernel="thin_plate",          # "thin_plate", "linear", "cubic"
    spline_order=3,
    corr_type="Subtraction",
    ai_path=None,
    ai_gpu_acceleration=False
)
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.