JEITA CP-3451
- 68 -
4.8 Data Description
4.8.1 Stipulations on Compressed Image Size
The JPEG format is adopted for compressed images. Since JPEG compresses images in blocks of 8 x 8 pixels,
preferably the pixel number set as the width and height of the image should be exact multiples of the MCU
(multiples of 16 for width, and multiples of 8 for height if Y:Cb:Cr = 4:2:2, or 16 if Y:Cb:Cr = 4:2:0). It is not required,
however, that all image data be fit within this range.
If a restart marker is inserted every 4 MCU for the sake of transformation to Flashpix, the width of the image shall
be an exact multiple of 4 MCU (64 pixels). There is no such restriction on height.
Specifications on inserting data to adjust the recorded image size relative to the valid image size, or padding, are
given below.
A. Image size recording procedure in the Exif writer
An Exif writer shall record a compressed image with the image width information and height information set in
accord with the stipulations below, and padded as necessary.
When the image is padded in the direction of width, the padding data shall be appended at the right edge. Padding
in the direction of height shall be appended at the bottom edge.
The data used for padding is not specified.
Stipulations on image width
Image width information consists of the valid image width tag (PixelXDimension), SOF marker information, and
JPEG compressed data. Of these three, the recording of a valid image width tag is mandatory.
Image width is padded to the right in the following cases. The value used for padding is not specified.
· When the image width is not an exact multiple of 16 and the JPEG block has unused capacity to be filled.
· When the image width is not a multiple of 64 and the insufficiency shall be made up in order to insert restart
markers.