We present an automated numerical method of compensating for retinal shadows in the choroid. In this method, signal extinction caused by retinal vessels is estimated by subtracting median A-scans obtained from beneath the retinal vessels and A-scans from the surrounding area. Adding the obtained offset vector to A-scans from beneath the retinal vessels allows compensating for shadows in the choroid. In vivo imaging of the human eye was performed by 840-nm-band standard resolution spectral domain optical coherence tomography (SD-OCT), and choroidal vasculature projection images were calculated. Removal of retinal shadows distinctly improved the readability of choroidal images.