The proposed H.261
based video codec is shown in Figure 2.
The adaptive quantizer is implemented using a VSLA. This will
be discussed subsequently. To prevent synchronization loss
due to error propagation in the variable length coded transmitted data
a error-resilience code called FEREC [10] is
used. We now discuss a channel matched source quantization scheme.
This is similar to the one proposed for the transmission
of JPEG compressed images [10].
The video frames are categorized into two classes, namely,
the intra frame or the I-frame and the predicted frame or the P-frame.
One in every 32 frames are intra frame coded. The I-frame coding
is similar to the JPEG still image coding which consists of
the discrete cosine transform (DCT), quantization and Huffman encoding.
The P-frame coding is based on DPCM and motion estimation. I-frames
are coded without reference to the preceding frames; whereas
the P-frames are coded with respect to the temporally closest preceding
I-frame. In P-frame coding the best match for each macroblock of the
current frame is found in a search area in the previous intra frame
using a block matching technique. The two macroblocks are subtracted and
the difference is transformed using DCT, quantized and Huffman encoded.
Motion estimation is done based on an exhaustive search based
block matching technique [4]. The quantization of the
DCT coefficients is implemented based on scaling each coefficient
by an entry in the quantization table [1].
We assume that the same quantization table is used
for I and P frames in this project for simplicity. However, the
proposed joint rate control and channel estimation is independent
of this assumption. The quantization table for DCT
blocks used for the simulations
is shown in Figure 3. Clearly, the DC coefficient and the
low frequency AC coefficients are finely quantized. The high frequency
AC coefficients which have less energy are coarsely quantized.
DCT
blocks used for the simulations
is shown in Figure 3. Clearly, the DC coefficient and the
low frequency AC coefficients are finely quantized. The high frequency
AC coefficients which have less energy are coarsely quantized.
