- class autoarray.inversion.pixelizations.VoronoiBrightnessImage(pixels=10, weight_floor: float = 0.0, weight_power: float = 0.0)#
- __init__(pixels=10, weight_floor: float = 0.0, weight_power: float = 0.0)#
A pixelization associates a 2D grid of (y,x) coordinates (which are expected to be aligned with a masked dataset) with a 2D grid of pixels. The Voronoi pixelization represents pixels as an irregular grid of Voronoi cells which can form any shape, size or tesselation.
Both of these grids (e.g. the masked dataset’s 2D grid and the grid of the Voronoi pixelization’s pixels) have (y,x) coordinates in two reference frames:
data: the original reference frame of the masked data.
source: a reference frame where grids in the data reference frame are transformed to a new reference
frame (e.g. their (y,x) coordinates may be shifted, stretched or have a more complicated operation performed on them).
The grid associated with the masked dataset and Voronoi pixelization have the following variable names:
grid_slim: the (y,x) grid of coordinates of the original masked data (which can be in the data frame and
given the variable name data_grid_slim or in the transformed source frame with the variable name source_grid_slim).
pixelization_grid: the (y,x) grid of Voronoi pixels which are associated with the grid_slim (y,x)
coordinates (association is always performed in the source reference frame).
A Voronoi pixelization has four grids associated with it: data_grid_slim, source_grid_slim, data_pixelization_grid and source_pixelization_grid.
If a transformation of coordinates is not applied, the data frame and source frames are identical.
Each (y,x) coordinate in the source_grid_slim is associated with the Voronoi pixel whose centre is its nearest neighbor. Voronoi pixelizations do not use a weighted interpolation scheme (unlike the Delaunay) pixelization.
For the VoronoiBrightnessImage pixelization the centres of the Voronoi grid are derived in the data frame, by applying a KMeans clustering algorithm to the masked data’s values. These values are use compute pixels number of pixels, where the weight_floor and weight_power allow the KMeans algorithm to adapt the derived pixel centre locations to the data’s brighest or faintest values.
In the project PyAutoLens, one’s data is a masked 2D image. Its data_grid_slim is a 2D grid where every (y,x) coordinate is aligned with the centre of every unmasked image pixel. A “lensing operation” transforms this grid of (y,x) coordinates from the data frame to a new grid of (y,x) coordinates in the source frame. The pixelization is then applied in the source frame.. In lensing terminology, the data frame is the image-plane and source frame the source-plane.
pixels – The total number of pixels in the Voronoi pixelization, which is therefore also the number of (y,x) coordinates computed via the KMeans clustering algorithm in data frame.
weight_floor – A parameter which reweights the data values the KMeans algorithm is applied too; as the floor increases more weight is applied to values with lower values thus allowing Voronoi pixels to be placed in these regions of the data.
weight_power – A parameter which reweights the data values the KMeans algorithm is applied too; as the power increases more weight is applied to values with higher values thus allowing Voronoi pixels to be placed in these regions of the data.
__init__([pixels, weight_floor, weight_power])
A pixelization associates a 2D grid of (y,x) coordinates (which are expected to be aligned with a masked dataset) with a 2D grid of pixels.
Computes the pixelization_grid in the data frame, by overlaying a uniform grid of coordinates over the masked 2D data (see Grid2DSparse.from_grid_and_unmasked_2d_grid_shape()).
Mapper objects describe the mappings between pixels in the masked 2D data and the pixels in a pixelization, in both the data and source frames.
Return the Voronoi source_pixelization_grid as a Grid2DVoronoi object, which provides additional functionality for performing operations that exploit the geometry of a Voronoi pixelization.
Relocates all coordinates of the input source_grid_slim that are outside of a border (which is defined by a grid of (y,x) coordinates) to the edge of this border.
Relocates all coordinates of the input source_pixelization_grid that are outside of a border (which is defined by a grid of (y,x) coordinates) to the edge of this border.
Computes a weight_map from an input hyper_image, where this image represents components in the masked 2d data in the data frame.