back Auction Sniper

chromatic aberration mitigation

Images of interest will typically be captured
by sensors generating nominally red, green and blue pixels.
Ideally, microscope calibration slide images would have
perfectly registered light squares in dark fields
with equal red, green and blue amplitudes for each pixel.
  • finite resolution has light squares more nearly circular blobs for some objectives
  • chromatic aberrations have red and blue blob centers misregistered from green
  • corresponding pixels have different red, green and blue amplitudes.
      - reduce color correction errors by histogram analyses:

    .. estimating corresponding blob edge values (see ellipses).
  • having identified threshold values to binarize blobs, then perform 2D blob alignments,
    assuming that misregistrations near image centers have overlapped blob centroids.
  • Using tabulated corresponding blob centroids as fiducials,
    perform non-rigid registration of red and blue images to green.
      - non-rigid image registration projects
      - optical flow - in C++
  • Save polynomial regression coefficients of centroid corrections
    as chromatic aberration profile for that objective

Image Registration Methods - nonrigid registration

maintained by blekenbleu