Image Processing

Image Processing#

vampires_dpp.image_processing.adaptive_sigma_clip_mask(data, sigma=10, boxsize=8)#

Compute a sigma-clip bad pixel mask using non-overlapping local blocks.

Parameters:

  • data (NDArray) – 2D image

  • sigma (float) – Sigma threshold for clipping

  • boxsize (int) – Size of the local block for computing statistics

Returns:

Boolean mask, True where pixels are clipped

Return type:

NDArray

vampires_dpp.image_processing.crop_to_nans_inds(data)#

Crop numpy array to min/max indices that have finite values. In other words, trims the edges off where everything is NaN.

Return type:

ndarray[tuple[int, ...], dtype[TypeVar(_ScalarType_co, bound= generic, covariant=True)]]

Parameters:

data (ndarray[tuple[int, ...], dtype[_ScalarType_co]])

vampires_dpp.image_processing.derotate_cube(data, angles, **kwargs)#

Derotate a cube clockwise frame-by-frame with the corresponding angle vector.

Parameters:

  • data (ArrayLike) – 3D cube to derotate

  • angles (ArrayLike | float) – If a vector, will derotate each frame by the corresponding angle. If a float, will derotate each frame by the same value.

Returns:

Derotated cube

Return type:

NDArray

vampires_dpp.image_processing.derotate_frame(data, angle, center=None, **kwargs)#

Rotate a single frame clockwise by the given angle in degrees.

Parameters:

  • data (ArrayLike) – 2D frame to derotate

  • angle (float) – Angle, in degrees

  • center (Optional[list | tuple]) – Point defining the axis of rotation. If None, will use the frame center. Default is None.

  • **kwargs – Keyword arguments are passed to warp_frame

Returns:

Derotated frame

Return type:

NDArray

vampires_dpp.image_processing.shift_cube(cube, shifts)#

Translate each frame in a cube using vectorized Fourier-domain shifting.

Processes the entire cube in a single FFT call with no Python loop. NaN values are filled per-frame before the FFT and the NaN boundary is restored by shifting the NaN mask through the same transform.

Parameters:

  • cube (ArrayLike) – 3D cube (nframes, ny, nx)

  • shifts (ArrayLike) – Array of (dy, dx) pairs, one for each frame in the input cube

Returns:

Shifted cube

Return type:

NDArray

vampires_dpp.image_processing.shift_frame(data, shift)#

Shift a single frame by the given offset using Fourier-domain shifting.

Supports sub-pixel shifts. Uses periodic boundary conditions. NaN values are filled before the FFT and the NaN boundary is restored by shifting the NaN mask through the same transform.

Parameters:

  • data (ArrayLike) – 2D frame to shift

  • shift (tuple[float, float]) – Shift (dy, dx) in pixels

Returns:

Shifted frame

Return type:

NDArray

vampires_dpp.image_processing.warp_frame(data, matrix, antialias=False, **kwargs)#

Geometric frame warping using Lanczos4 interpolation with NaN padding by default.

Parameters:

  • data (ArrayLike) – 2D image

  • matrix (ArrayLike) – Geometric transformation matrix

  • antialias (bool) – Apply Gaussian blur before warping to reduce aliasing when downsampling.

  • **kwargs – Keyword arguments passed to opencv (e.g. borderValue, borderMode).

Returns:

Warped frame

Return type:

NDArray

Contents