FIG. 18 shows a macroblock decoding process including a step of receiving a flag indicating whether to perform intra prediction on a sub-macroblock according to an embodiment of the present invention. If an intra prediction type is added to a sub-macroblock, as the number of sub-macroblock types is incremented, a data size to be transmitted may increase. To solve this problem, by transmitting a flag (sub_mb_intra_pred_flag) confirming whether a sub-macroblock performs intra prediction before transmitting a sub-macroblock type, if the flag is set to 1, a block size and an intra prediction mode are transmitted instead of a sub-macroblock type. On the contrary, if the flag is set to 0, a sub-macroblock type is transmitted. And, a macroblock is decoded using a received type and motion information.

In H.264/AVC, an analog video is parsed into a higher frequency component and a lower frequency component using orthogonal transform of 8×8 or 4×4 pixel unit to facilitate compression. If a size of a macroblock becomes greater than 16×16, it is necessary to raise compression efficiency using a proper transform block size. Although a 4×4 or 8×8 transform block size used for H.264/AVC may be available, it may be necessary to perform a transform having a greater block size to secure higher compression efficiency.