Vignetting correction by exploiting an optical microscopy image sequence.

Vignetting is one of the most common problem that may affect digital imaging. The effect becomes particularly evident when images are stitched together to increase the camera's field of view (e.g., when building a mosaic), where it can lead to errors in automatic analyses. To correct the effect, the most common approach is to acquire an empty field image in advance that is used later to perform a flat field correction on every subsequently acquired image. However, in several cases, such as when dealing with off-line images or with real time acquisitions, this is not a viable option. The method we propose relies on a non parametric model to characterize in real time the vignetting function from the specimen itself, by using our foreground/background segmentation algorithm. The function is computed over a background built incrementally, detecting regions free of objects of interest. The experiments carried out using cell cultures and histological samples prove that our method yields results at least comparable to those achieved by using empty field.