bookmark CIDRE / corrected intensity distributions using regularized energy minimization A retrospective illumination correction method for optical microscopy. CIDRE is designed to correct collections of images by building a model of the illumination distortion directly from the image data. Larger image collections provide more robust corrections. CIDRE achieves correction quality similar to that of prospective methods. Unlike existing retrospective methods, CIDRE estimates both linear intensity gain function and an additive term. The key insight to our approach is that the distribution of intensities from a single location across many images is related to an underlying distribution common to all locations by a linear transform. This assumes that objects may appear anywhere in the image with equal probability.