An image processing apparatus for correcting hand shake blur in a moving image includes a generation unit that generates, by setting an image among images constituting the moving image as a reference image and the remaining images as correction target images, a correction amount by which position displacement of a correction target image relative to the reference image is corrected; an estimation unit that estimates a hand shake blur area where hand shake blur relative to the reference image has occurred, on the basis of the correction amount; a detection unit that detects a composition changing operation on the basis of the correction amount; and an updating unit that updates a correction area to an area where hand shake blur in the correction target image relative to the reference image is corrected on the basis of the hand shake blur area and a detection result.