changes_from_vm3_0a.txt

来自「JPEG2000实现的源码」· 文本 代码 · 共 53 行

                         CHANGES IN VM3.1A FROM VM3.0A
	                 =============================

          Author: David Taubman
          Affiliation: Hewlett-Packard Research Labs, California, USA
                  and  The University of New South Wales, Sydney, Australia
          Email: [e-mail]

______________________________________________________________________________
                                 New Features
______________________________________________________________________________

Practically all capabilities which were missing from VM3.0A have been
included in VM3.1A, plus a number of unexpected extras.  The new features
are:

* Introduction of frames as a way of supporting the ideas behind the SSWT
  in a general and uniform way.

* Memory efficient arbitrary ROI mask generation, with circles and
  rectangles implemented as special cases.

* Progressive visual weighting implemented.

* Memory accounting, reporting and leak tracing capabilities.

* Replacement of all "local_printf()"/"local_exit()" combinations with
  "local_error()", to ensure consistent and convenient error reporting.

______________________________________________________________________________
                                  Bug Fixes
______________________________________________________________________________

* Bug in the distortion computation code for EBCOT in the LSB of
  lossless quantization indices has been fixed -- results in better
  performance at high bit-rates in progressive-to-lossless applications.

* Source of some warning messages from Solaris compiler eliminated.

______________________________________________________________________________
                                 More Results
______________________________________________________________________________

* Results reported in "vm3a_performance.xls" have been modified to
  reflect bug fix mentioned above.

* Results extended to the following images:
     "Aerial2", "Barbara", "Bike", "Cafe", "Chart", "Compound 2",
	 "CR", "Finger", "Gold", "Hotel", "Lenna", "SAR 1", "SAR 2",
	 "Tools", "Texture 1", "Texture 2", "Woman" and "X-Ray"
  with all eight test conditions described in READ_THIS_FIRST.TXT applied
  to each image.