|
|
|
Usuarios conectados
Actualmente hay 5647 visitantes online. |
|
Productos
|
|
Información
|
|
Destacado
|
|
|
 |
|
|
No hay comentarios de productos.
FAX-L350 Chapter 2: Technical Reference
b) In the model template, the combination (10-bit pixel pattern) of 10 pixels is output to the arithmetic encoding section using the template shown below (inside the bold outline). All of the 10-bit pixel patterns inside this template exist in the study table. This 10-bit pixel pattern is used by the arithmetic encoding section to refer to the predicted value of the pixel and the status number in the study table which correspond to the 10-bit pixel pattern.
AT X X
AT
3-line model template
2-line model template
Figure 2-18 Model Templates
There are two types of model templates 3-line and 2-line, and the one selected is designated by the LRLTWO inside the Bi-level Image Header section (BIH). The pixel shown by �X� is an encoded pixel and is outside of the template. The pixel shown by �AT� is a special pixel known as an AT pixel. The AT pixel becomes a Adaptive Template by having its position moved, and is very effective when encoding a periodic pixel, similar to a dither pattern image. The position of AT in the figure is the beginning position of the AT pixel. In this fax, it remains fixed in this initial position, so Adaptive Template is not used.
NOTE
c) The study table, as shown below, is constructed by all of the 10-bit pixel patterns output by the model template, and their corresponding status numbers and predicted values of the pixel to be encoded. The predicted value of the pixel to be encoded and the status number is compared to the actual pixel in the arithmetic encoding section every time the model template is moved to the adjoining pixel. The result of this comparison (matches / does not match predicted value) and the status number are then checked by comparison to the probability estimation table, and the study table is corrected (learned) to a new prediction value and status number which will be used when the same pixel pattern is found again. By learning in this way, the probability of the study table matching the next time is increased, and the need for encoding decreased.
2-41
|
|
|
> |
|
