Statistics

Basic image statistics and histogram analysis. All functions operate on &[f32] slices with no heap allocation.

Basic Statistics

compute returns a Stats struct from a data slice:

Field	Description
count	Number of samples
min	Minimum value
max	Maximum value
mean	Arithmetic mean
variance	Population variance

Accumulation uses f64 internally to reduce floating-point error on large arrays.

Histogram

histogram bins a data slice into n_bins uniform bins over a caller-specified [min, max] range. The output is a &mut [u32] slice whose length determines the number of bins. Values at or above max are placed in the last bin.

let mut bins = [0u32; 256];
histogram(&data, 0.0, 1.0, &mut bins);

Percentiles

percentile_from_histogram estimates a percentile value from a pre-computed histogram. The percentile parameter is in [0.0, 1.0]. It walks the cumulative distribution and returns the bin center where the target count is reached.

let median = percentile_from_histogram(&bins, 0.0, 1.0, 0.5);
let p95 = percentile_from_histogram(&bins, 0.0, 1.0, 0.95);

Resolution is limited by the number of bins. For 256 bins over [0, 1], the resolution is ~0.004.