To always obtain compressed image data with corrected image quality having excellent image quality from compressed image data.

A correction amount computing means 15 computes correction amounts of white balance correction and density correction which are to be applied to image data D1 obtained by decompressing compressed image data P1 by a first decompression means 10, and an image quality correction means 25 applies image quality processing to the image data D1 by using these correction amounts to obtain the image data D2 whose image quality is corrected. A second decompression 35 decompresses the compressed image data P2 obtained by applying compression processing to the data D2 by a compression means 30 to obtain RGB image data D3. A first change amount computing means 20 and a second change amount computing means 40 respectively compute a signal change amount S1 obtained by image quality correction executed by an image quality correction means 25, and a signal change amount S2 obtained by compression processing executed by the means 30. An object image data deciding means decides the data P2 as object image data P if the amount S1 is larger than S2, and decides the original data P1 as the data P if S1 is equal to or lower than S2.