Atmel Evaluation Kit AT91SAM9M10-G45-EK AT91SAM9M10-G45-EK Data Sheet

 1110

SAM9M10 [DATASHEET]

6355F–ATARM–12-Mar-13

Notes:

1. Step size in the start and end coordinates depends on the output picture format and the width of the data bus.

2. Usable in stand-alone post-processing mode only.

The decoder software starts decoding the first picture by parsing the stream headers (

). Software then setups the

hardware control registers (picture size, stream start address etc.) and enables the hardware (

).

In JPEG case, software parses VLC and QP tables from the stream and writes them to a contiguous memory area.
In VC-1 case, software parses bitplane macroblock control from the stream and writes it to memory.

Hardware decodes the picture by reading stream, VLC and QP tables for JPEG (

), and the reference pictures (

)

(required for inter picture decoding) from the external memory. Hardware writes the decoded output picture to
memory (

) one macroblock at a time. When the picture has been fully decoded, or the hardware has run out of

stream data, it gives an interrupt with a proper status flag and provides stream end address for software to con-
tinue and returns to initial state.

In case the hardware detects erroneous stream during decoding, it will return an error interrupt, and software will
start concealing errors for the current picture. Depending on the selected concealment method, software may re-
initialize the hardware in the SW entropy decode mode. The pictures following the corrupted one are decoded in
the normal mode until the next error is detected.

If post-processing is enabled, one or two additional image transfer operations will take place. If the decoded
images are in display order (i.e. no picture re-ordering has been made when encoding the sequence), and rotation
is not used, PP will process the pictures in pipeline with the decoder. Otherwise, it will first have to read the
decoded image that is to be displayed next from the memory (

), and then write back the processed image (

).

Linear

Using a modified H.264 in-loop filter as a post-processing filter. Filtering 
has to be performed in combined mode.

User definable start position, height and width. Can be used with scaling to 
perform digital zoom. Usable only for JPEG or standalone mode.

Output image can be written to any location inside video memory 

Up to 1280 x 720

Output image writing can be prevented on two rectangular areas in the 
image 

however it 

is possible to have one masking area and one blending area.

Rotation 90, 180 or 270 degrees

Horizontal flip

Vertical flip

Postprocessing Features (Continued)