A support mask decides which pixels are eligible for segmentation. It is not a segmentation map.
Starlet support
The default path collapses the cube, reconstructs selected starlet scales, and thresholds the result.
support <- build_starlet_mask(
input = cube,
starlet_J = 5,
starlet_scales = 2:5,
denoise_k = 2.5,
positive_only = TRUE
)
image(support$mask)The support depends on the collapse function, scale range, threshold, and cleanup settings.
Adaptive support
build_adaptive_support() combines evidence across bands.
It is useful when no single collapse represents the source
adequately.
support <- build_adaptive_support(cube, pretransform = "none")For directional structures, build_contourlet_mask()
provides an alternative to starlets. Compare both masks before
segmentation.