Channel coding performance of an adaptive DCT image coding in combination with m‐transform

In the conventional adaptive discrete cosine transform (ADCT) using the fixed length code, it is known that the source coding performance can be improved greatly by the combined use of the M‐transform (MT). This method is called ADCT‐MT. The purpose of this paper is to evaluate the effect of the combined use of MT from the viewpoint of the channel coding performance. In the conventional ADCT, as well as in the scene adaptive coding (SAC) using the variable length code, the damage in the reconstructed image due to the channel error often is manifested as the local block distortion. In ADCT‐MT, the damage is diffused and distributed uniformly by the inverse MT over a large number of blocks, making the damage less noticeable. A subjective evaluation experiment is conducted (five‐step score) assuming the binary symmetrical channel. A simple error‐correcting code is applied to the coding data of the additional information and DCT dc component. When the bit‐error rate is 10⁻³, the average score is approximately 3.5 for ADCT and 1.2 for SAC, while it is above 4.5 in ADCT‐MT. Thus, in ADCT‐MT, the damage is little noticeable.