An image processing device includes a first image acquisition unit that acquires each of a first imaging signal indicating a flash emission image and a second imaging signal indicating a flash non-emission image, a second image acquisition unit that acquires a third imaging signal indicating a difference between the first imaging signal and the second imaging signal that have been acquired, a third image acquisition unit that acquires a fourth imaging signal obtained by multiplying the acquired third imaging signal by a white balance gain for a flash for removing an influence due to color of the flash light, a color signal acquisition unit that acquires a color signal indicating a first color of each area in an imaging screen on the basis of the acquired fourth imaging signal, and a first white balance gain calculation unit.