|
| Hi,
Based on the guidance of Gurus in this group, I just started my
implementation
of real time image compression using JPEG-LS.
Please don't frown upon me, I know my question is so obvious.. :-)
As defined in stanadard specifications "FCD14495 (Public Draft)" Annex
H,
I went through all the process of encoding by hand computations and
generated
the resulting stream as given in the specs.
While working with the decoding process, I initialized the variables
from the
configuration data of encoded stream but I am struck in the
computation of local
gradients, (a very first step after initialization). I went through
the implementation
from University of British Columbia but the implementation seems to be
processing
data with multiple steps merged into one function.
I need to know:
1- How the scan lines (Previous and Current Scan Line for decoding)
are formed from an encoded stream like "C0 00 00 6C 80 20 8E 01 C0 00
57 ....."; which in turn form the causal template for computation of
gradients?
2- An example of encoding is given in the specs (FCD14495), which made
it very easy to understand the encoding process. Is there any example
of decoding on small image (elaborating all the steps), BTW I could
not find any ?
/MH
|
|