An image processing apparatus and method decodes interceded image data, even if reference image data is lost. The apparatus and method perform first coding that codes input image data by transforming the input image data into frequency components in units of blocks and coding the frequency components by adaptively using an intracoding mode and an interceding mode, second coding that codes frequency components obtained by limiting the frequency components for reference image data for use in the interceding mode, and outputting combined data obtained by combining the image data coded in the second coding with the image data coded by the first coding.