Sharpness is calculated in all of focus-bracketed images on a pixel basis. Then, a first reference value indicating an image of the plurality of images to which a pixel whose sharpness is the highest among the pixels located on the identical positions in the plurality of images belongs is obtained on each pixel of the images, and a second reference value is calculated based on the first reference value on each pixel by spatially smoothing the first reference value on each pixel based on the first reference values on adjacent pixels. The focus-bracketed images are processed based on the second reference values to generate an omni-focus image or a blur-enhanced image. Accordingly, it is possible to judge a region having high contrast as an in-focus region and acquire a synthesized image having smooth gradation.