The DREV98 image format was developed to conduct a number of experiments on the transfer of metadata between an auto-documented sensor and an auto-loading image database. These experiments allowed us to determine which metadata had to be stored in the image header. Moreover, this image format brings the concept that the header segments should contain meta-metadata, ie. information that describes the header segment data (which itself describes the image) allowing a universal decoder to decode all unknown segment. This technical note contains the description of the DREV98 image format and the new concept of the self-documented header segments.