scarlet2.detect

Detection methods

Higher-level helpers (get_wavelets, QuadTreeRegion, get_peaks, …) are adapted from scarlet1’s detect.py.

Uses NumPy rather than JAX because detection involves dynamic data structures (variable-length peak lists, irregularly shaped footprints) that are not compatible with JAX’s JIT. Both NumPy and JAX arrays are accepted as inputs.

Classes

Footprint(footprint, peaks, bounds)	A detected footprint (connected region above threshold) in an image.
HierarchicalFootprint(peak, int], bbox, ...)	A source detected in the starlet hierarchy.
Peak(y, x, flux)	A peak (local maximum) in a Footprint.
QuadTreeRegion(bbox[, capacity, ...])	A QuadTree that stores bounding boxes (rather than points).
SingleScaleStructure(scale, footprint)	A connected set of pixels with common peaks at a single wavelet scale.

Functions

box_intersect(box1, box2)	Check whether two Box instances overlap.
footprint_intersect(footprint1, box1, ...)	Check whether two footprint masks overlap.
footprint_iou(source1, source2)	Compute intersection over union (IoU) between two source footprints.
footprints(image[, min_separation, ...])	Detect footprints and their peaks in an image.
get_blend_structures(detect[, scales, ...])	Build SingleScaleStructure objects for the third wavelet scale.
get_blend_trees(detect[, scales, ...])	Build a QuadTreeRegion for each wavelet scale in detect.
get_detect_wavelets(image, variance[, ...])	Get starlet coefficients of a detection image for source finding.
get_wavelets(images, variance[, max_scale])	Compute significant starlet coefficients for a multi-band image cube.
hierarchical_footprints(obs[, scales, ...])	Decompose an observed image into a list of HierarchicalFootprint objects.
multiscale_footprints(obs[, scales, strict, ...])	Detect footprints at multiple starlet scales from an observation.
split_footprint(fp, image[, min_area])	Split a multi-peak Footprint into single-peak sub-footprints.