Histogram Matching

Updated 02 Oct 2026

•
What it is
- •
  also claled histogram specification: transforms an image so its intensity distribution resembles a chosen reference distribution
•
Mechanism
- •
  map suorce intensities through their cumulative distribution and then through an inverse reference cumulative distribution: $z=F_{\mathrm{ref}}^{-1}(F_{\mathrm{src}}(r))$ .
- •
  discrete implementations approximate this mapping
•
Color images
- •
  match_histograms(..., channel_axis=-1) matches corresponding channels independently.
- •
  matching separate channel distributions does not guarantee matching the joint color distribution.
•
Distinctions
- •
  source and reference images need compatible channels, but not identical dimensions.
- •
  matching histograms does not align objects or transfer spatial arrangement.
•
Reference
- •
  https://docs.opencv.org/4.4.0/d0/d86/tutorial_py_image_arithmetics.html
•
Related
- •
  Image Histogram
- •
  Histogram Equalization

Linked references 3