viz.enhance.NormStats

viz.enhance.NormStats(
    radius,
    statistic,
    std,
    n_samples,
    n_valid,
    n_used,
    sampling,
    bin_edges=None,
)

Describe normalization estimates and support for each radial bin.

This record stores values from one enhancement pass; it performs no estimation. Its estimate and count arrays are one-dimensional, with one entry per polar ring or direct annular bin.

Attributes

radius : (ndarray, shape(n_bins))

Integer ring radii for polar sampling, or bin-edge midpoints for direct annuli, in pixels.

statistic : (ndarray, shape(n_bins))

Retained-sample divisor or location, before image transformation. Zero/negative values are reported even if unsuitable for division.

std : (ndarray, shape(n_bins))

Retained sample standard deviation (ddof=1), regardless of the clipping scale. One sample has zero spread; empty bins have NaN.

n_samples : (ndarray, shape(n_bins))

Total sampling opportunities. Polar rings have ntheta angular cells. Direct annuli count all in-frame pixel centers in each bin, including masked/nonfinite pixels; off-image area is not counted.

n_valid : (ndarray, shape(n_bins))

Finite samples after both masks, before sigma clipping.

n_used : (ndarray, shape(n_bins))

Samples retained after clipping. A callable may use these samples internally in its own way; that does not change this count.

sampling : {polar, pixel_center}

Whether counts refer to polar cells or original pixel centers.

bin_edges : (ndarray, shape(n_bins + 1) or None)

Exact direct-annulus edges in pixels. None for integer polar rings.

Notes

Counts satisfy 0 <= n_used <= n_valid <= n_samples. They describe estimation support, not valid output pixels. In polar sampling, repeated lookups of one source pixel are not independent measurements.

The arrays belong to the result and can be edited without changing the input image. Only n_valid and n_used exclude hard and estimation masks; n_samples counts opportunities before masking. n_rejected counts sigma rejections only.

Use astroimred.phot.radprof.radial_profile to calculate a standalone annular profile with sky_fit options. Matching statistics require matching sample selection, clipping, and degrees of freedom.