readme.binjpeg

JPEG Image compression using IJG standards followed

********************************************
	Readme File for binDCT-based JPEG
********************************************

1. Contact Information
-----------------------

% Trac D. Tran and Jie Liang
% ECE Department, The Johns Hopkins University
% 3400 North Charles Street, 105 Barton Hall,
% Baltimore, MD 21218
% E-mail: [e-mail], [e-mail]
% Dec. 2000

2. binDCT-related Files
-----------------------

This implementation of the binDCT is modified from IJG's package.
For conditions of distribution and use, see the accompanying README file.

Encoder side:
------------
jfdct_bin_a1.c
	Forward binDCT_a1: 9  Shifts, 28 Adds. Coding gain: 8.7686dB.
	This is the Version C7 in the IEEE TSP paper (because of relabelling).

jfdct_bin_b1.c
	Forward binDCT_b1: 19 Shifts, 37 Adds. Coding gain: 8.8220dB.
	This is the Version C4 in the IEEE TSP paper.

jfdct_bin_c1.c
	Forward binDCT_c1: 23 Shifts, 42 Adds. Coding gain: 8.8251dB.
	This is the Version C1 in the IEEE TSP paper.

jfdct_bin_l1.c
	Forward binDCT_l1: 16 Shifts, 34 Adds. Coding gain: 8.8225dB.
	Same as binDCT-L3 in the paper.

Decoder side:
-------------

jidct_bin_a1.c, jidct_bin_b1.c, jidct_bin_c1.c, jidct_bin_l1.c:
	Inverse transforms of the above.

4. BinDCT-related Options
-------------------------

  -dct int       Use integer DCT method (default)
  -dct fast      Use fast integer DCT (less accurate)
  -dct float     Use floating-point DCT method
  -dct bin_a1    Use binDCT-A1: 9 Shifts, 28 Adds. Coding gain: 8.7686dB.
  -dct bin_b1    Use binDCT-B1: 19 Shifts, 37 Adds. Coding gain: 8.8220dB.
  -dct bin_c1    Use binDCT-C1: 23 Shifts, 42 Adds. Coding gain: 8.8251dB.
  -dct bin_l1    Use binDCT-L1: 16 Shifts, 34 Adds. Coding gain: 8.8225dB.
  -lossless      Use lossless binDCT compression. 
                 No quantization and binDCT final scaling.
                 Only valid for binDCT algorithms. 
                 The same binDCT algorithm should be used in both encoding and decoding.


5. Examples
------------------

5.1 Using the original DCT (-dct int, -dct fast, or -dct float):

./cjpeg -quality 80 -dct int lena.pgm > lena_int.jpg
./djpeg -dct int lena_int.jpg > lena_int_rec.pgm

5.2 Using lossy binDCT (-dct bin_a1, -dct bin_b1, -dct bin_c1, or -dct bin_l1):

./cjpeg -quality 80 -dct bin_b1 lena.pgm > lena_b1.jpg
./djpeg -dct bin_b1 lena_b1.jpg > lena_b1_rec.pgm

5.3 Lossless binDCT: use -lossless at both sides

./cjpeg -dct bin_b1 -lossless lena.pgm > lena_b1.jpg
./djpeg -dct bin_b1 -lossless lena_b1.jpg > lena_b1_rec.pgm

6. Copyright

% - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
% Copyright (c) 2000 Trac D Tran and Jie Liang
% The binDCT-related program in this package is Copyright (c) 
% by Trac D Tran and Jie Liang.
% It may not be redistributed without the consent of the copyright
% holders. In no circumstances may the copyright notice be removed.
% The program may not be sold for profit nor may they be incorporated
% in commercial programs without the written permission of the copyright
% holders. This program is provided as is, without any express or
% implied warranty, without even the warranty of fitness for a
% particular purpose.
%-----------------------------------------------------------------------