Correction of photomicrographic radial distortion using a virtual lens

Abstract

A “virtual lens” which utilizes digital image processing is introduced as a flexible solution for correcting of radial distortion in photomicrography. In the first series of tests, photomicrographs taken from a hemacytometer with regular divided lines was used. A polynomial function was used to correct the digital image. We used Newton’s method to find an approximate root of the polynomial equation and derived the coordinate transform equation. The distortion error was measured from the mismatching shift distance of two sets of superimposed reference points. The optimal correction parameters were found by a two-stage search strategy. With the optimal correcting parameter used, the radial distortion of the grid was almost fully corrected. Before correction, the distortion errors of images taken with 5x and l0x objectives were 11.0±8.3 and 6.0±1.9 pixel/point, respectively. After correction, the distortion errors were reduced to 2.1± 0.7 and 1.2± 0.3 pixel/point. In the second series of tests, photomicrographs of actual histological sections were used. The different extent of distortion between images was corrected. The corrected images could be easily stitched together to become a higher-resolution image.