This invention makes it possible to detect only a relatively large mismatch while allowing a mismatch to some extent between an LDR image and HDR information, which are stored in a single file, caused by editing of the LDR image. An apparatus of this invention generates, from an input HDR image, an LDR image and difference information representing a difference between the LDR image and the HDR image. The generated LDR image is encoded. Information formed from data representing a local feature of the LDR image is calculated as LDR meta-data. The apparatus generates a file by storing the encoded data in a main encoded data storage region and storing the difference information and the LDR meta-data in a marker defined as a non-reference region for an LDR image decoding apparatus in the file structure.