This paper proposes a technique which can improve subjective picture q
uality for Video codec. On this approach, we focus on the importance o
f facial pictures in videophone and videoconferencing applications. Th
is algorithm is roughly divided into encoding and decoding processes.
For the encoding process, we classify the picture into the facial regi
on and background region. A low-pass filter is then adopted to elimina
te the high frequency components in the background region. The facial
region is preferentially allocated the number of bits by coding contro
l. For the decoding process, post-filtering for the reduction of mosqu
ito noise and blocking effect is applied to the reproduced picture. Si
mulations of 2 videophone sequences have been carried out. The MOS for
the reproduced pictures using the proposed technique increases about
40.0 - 63.9 %. From the results, the effectiveness of the proposed tec
hnique for video coding has been definitely proved.