A weighting coefficient sequence defining unit generates an array in which weighting coefficients are held at positions defined by the directional components of blur vectors, and adjusts the weighting coefficients to increase the sum of the frequency components of the array or reduce variations of the frequency components. A weighting unit multiplies respective captured images by corresponding weighting coefficients, generating L weighted captured images. The weighting unit then generates the synthesized image of the L weighted captured images. A corrected image generation unit performs deconvolution processing using the frequency component of the array in which the weighting coefficients are held at positions defined by the directional components of blur vectors, and the frequency component of the synthesized image. The corrected image generation unit performs an inverse frequency transform for the deconvolution processing result, generating an output image.