Thu Sep 9 15:32:07 1999 Eero Simoncelli * epic.c / collapse_ortho_pyr.c: - Fixed scaling of 9-tap QMF filters (they had been modified in epic.c) - Added LeGall's biorthogonal set (9-7), which is used in SPIHT. Mon Jul 14 10:25:27 1997 Eero Simoncelli * run_length.c: added check for block ending in zero-length counting loop. Thanks to pilot1@orca.vub.ac.be (George Destree) for reporting this. Fri May 2 10:08:01 1997 Eero Simoncelli * epic.c: Added explicit instructions for using the orthonormal filter set. Sat Aug 31 12:41:56 1996 Eero Simoncelli * Included collapse_pyr code from Rob Buccigrossi, modified to work with (nearly) orthonormal symmetric 9-tap QMF filters. See epic.c. Fri Oct 20 19:25:20 1995 Eero Simoncelli * Added warning for images with dimensions d that are NOT multiples of 2^(floor(log2(d/15))). Sun Oct 15 13:07:04 1995 Eero Simoncelli * Speedup to unepic: In collapse_pyr, don't collapse zero coefficients. * changed scale_factor to 128: binsize=0 now produces something closer to perfect reconstruction Tue Oct 10 15:22:13 1995 Eero Simoncelli * Changed test_image to be a PGM file. * Version 1.1 tarfile made Sun Oct 8 15:22:57 1995 Eero Simoncelli * NON-SQUARE IMAGES ARE NOW HANDLED (added butch@grip.cis.upenn.edu's mods to collapse_pyr). * PGM FILES ARE NOW READ (BUT NOT WRITTEN) (added butch's mods for reading PGM images). * Slight mods to makefile: use $*.c instead of $< for denoting target Tue Sep 19 18:40:57 1995 Eero Simoncelli * New file "USAGE" gives usage. * quantize.c: minimal binsize is now set based on the maximum abs value of pixels, rather than the range. Mon Jun 7 20:25:07 1993 Eero Simoncelli * Minor change: fixed up some error messages. * Added new internal_tranpose and internal_int_transpose that perform in-place transpose on non-square matrices. This removes all dependencies on square images except for collapse_pyr, which needs to be rewritten. July 1989 Eero Simoncelli * Original source code released.