next up previous contents
Next: Other Techniques and Current Up: Block Transform Coding Previous: Quantisation

Encoding

Further compression can be achieved by exploiting the statistical redundancy in the resulting quantised DCT coefficients. This can be achieved with run-length or entropy encoding, such as Huffman encoding. The general idea behind these encoders is that the most frequently occurring values are assigned shorter codes, with longer codes representing the less frequent values.

Particularly, for DCT coefficients, it is better to select the input to the run-length encoding by applying a special zig-zag block scanning pattern, as shown in Figure 2.4. This scanning ensures that the correlation of adjacent samples will be maximal, increasing the overall coding efficiency.


  
Figure 2.4: The zig-zag scanning pattern for the DCT coefficients.
\begin{figure}\centering\epsfig{file=zigzag.eps,width=3.2in}\end{figure}



Isaac Kokkinidis
1998-08-27
Hosted by www.Geocities.ws

1