jpgencoderdataunit.cpp

Jpeg编解码器的源代码

//
// Copyright (c) 1997,1998 Colosseum Builders, Inc.
// All rights reserved.
//
// Colosseum Builders, Inc. makes no warranty, expressed or implied
// with regards to this software. It is provided as is.
//
// See the README.TXT file that came with this software for restrictions
// on the use and redistribution of this file or send E-mail to
// [e-mail]
//

//
// JPEG Encoder Library.
//
// Title:   EncoderDataUnit Class Implementation
//
// Author: John M. Miano  [e-mail]
//
//

#include <cmath>
#include "jpgencoderdataunit.h"
#include "vcl.h"
using namespace std ;
namespace ColosseumPrivate
{

const double PI = acos (-1.0) ;

const double FC4 = cos (PI * 4.0 / 16.0) ;
const double FSEC2 = 0.5 / cos (PI * 2.0 / 16.0) ;
const double FSEC6 = 0.5 / cos (PI * 6.0 / 16.0) ;


//
//  Description:
//
//    This is an implementation of the Forward Discrete Transform based
//    on matrix factorization of the DCT matrix. My first factorization
//    always left a constant factor of 1/8 at the end which could be merged
//    with quantization or as part of an integer descaling.
//
//    Using the cosine product formula it was possible to eliminate some
//    multiplication operations which resulted in a more complex scaling
//    matrix.
//
//    I have documented the derivation process for this DCT process as well
//    as another I have tried. I have some more factorization ideas to try
//    out when I have to to get around to it. Unfortunately matrix factorization
//    is a very tedious process. When you see the documents it looks easy, but
//    believe me and all the legal pad I went through that it is not.
//
//    This implementation is a litteral mapping of the matrix implementation.
//    Each set of temporaries represents one matrix multiplication. Hopefully
//    your compile will optimize the temporaries out. It is possible to
//    reorder the operations to reduce the number of temporaries variables
//    required (The Intel C++ compile does this on its own) which seems to be
//    the best optimization to try next.
//
//
//    I have not implemented a scaled integer version of the FDCT because I
//    believe that most people will want quality over speed in encoding.
//
//  Parameters:
//    qt:  The quantization table
//    output:  The output DCT coefficients
//

void JpegEncoderDataUnit::ForwardDct (JpegEncoderQuantizationTable &qt,
                                      JpegCoefficientBlock &output)
{
  double tmp [JPEGSAMPLEWIDTH][JPEGSAMPLEWIDTH] ;

  for (unsigned int col = 0 ; col < JPEGSAMPLEWIDTH ; ++ col)
  {
    // Shift this first matrix multiplication includes the range shift from
    // 0...255 to -128...127. Notice that the shift term is canceled out
    // in the alst four elements.
    double a0 = data [0][col] + data [7][col] - 2 * JPEG8BITMIDPOINTSAMPLEVALUE ;
    double a1 = data [1][col] + data [6][col] - 2 * JPEG8BITMIDPOINTSAMPLEVALUE ;
    double a2 = data [2][col] + data [5][col] - 2 * JPEG8BITMIDPOINTSAMPLEVALUE ;
    double a3 = data [3][col] + data [4][col] - 2 * JPEG8BITMIDPOINTSAMPLEVALUE ;
    double a4 = data [3][col] - data [4][col] ;
    double a5 = data [2][col] - data [5][col] ;
    double a6 = data [1][col] - data [6][col] ;
    double a7 = data [0][col] - data [7][col] ;

    double b0 = a0 + a3 ;
    double b1 = a1 + a2 ;
    double b2 = a1 - a2 ;
    double b3 = a0 - a3 ;
    double b4 = a4 ;
    double b5 = a5 ;
    double b6 = a6 ;
    double b7 = a7 ;

    double c0 = b0 ;
    double c1 = b1 ;
    double c2 = b2 + b3 ;
    double c3 = b3 ;
    double c4 = b4 + b5 ;
    double c5 = b5 + b6 ;
    double c6 = b6 + b7 ;
    double c7 = b7 ;

    double d0  = c0 ;
    double d1  = c1 ;
    double d2  = c2 ;
    double d3  = c3 ;
    double d4  = c4 + c6 ;
    double d5  = c5 ;
    double d6  = c6 ;
    double d7  = c7 ;

    double e0 = d0 + d1 ;
    double e1 = d0 - d1 ;
    double e2 = FC4 * d2 ;
    double e3 = d3 ;
    double e4 = FC4 * d4 ;
    double e5 = FC4 * d5 ;
    double e6 = d6 ;
    double e7 = d7 ;

    double f0 = e0 ;
    double f1 = e1 ;
    double f2 = e2 ;
    double f3 = e3 ;
    double f4 = e4 + e6 ;
    double f5 = e5 ;
    double f6 = e4 - e6 ;
    double f7 = e7 ;

    double g0 = f0 ;
    double g1 = f1 ;
    double g2 = f2 ;
    double g3 = f3 ;
    double g4 = FSEC2 * f4 ;
    double g5 = f7 - f5 ;
    double g6 = FSEC6 * f6 ;
    double g7 = f5 + f7 ;

    double h0 = g0 ;
    double h1 = g1 ;
    double h2 = g2 + g3 ;
    double h3 = g3 - g2 ;
    double h4 = g4 + g7 ;
    double h5 = g5 + g6 ;
    double h6 = g5 - g6 ;
    double h7 = g7 - g4 ;

    tmp [0][col] = h0 ;
    tmp [1][col] = h4 ;
    tmp [2][col] = h2 ;
    tmp [3][col] = h6 ;
    tmp [4][col] = h1 ;
    tmp [5][col] = h5 ;
    tmp [6][col] = h3 ;
    tmp [7][col] = h7 ;
  }

  for (unsigned int row = 0 ; row < JPEGSAMPLEWIDTH ; ++ row)
  {
    double a0 = tmp [row][0] + tmp [row][7] ;
    double a1 = tmp [row][1] + tmp [row][6] ;
    double a2 = tmp [row][2] + tmp [row][5] ;
    double a3 = tmp [row][3] + tmp [row][4] ;
    double a4 = tmp [row][3] - tmp [row][4] ;
    double a5 = tmp [row][2] - tmp [row][5] ;
    double a6 = tmp [row][1] - tmp [row][6] ;
    double a7 = tmp [row][0] - tmp [row][7] ;

    double b0 = a0 + a3 ;
    double b1 = a1 + a2 ;
    double b2 = a1 - a2 ;
    double b3 = a0 - a3 ;
    double b4 = a4 ;
    double b5 = a5 ;
    double b6 = a6 ;
    double b7 = a7 ;

    double c0 = b0 ;
    double c1 = b1 ;
    double c2 = b2 + b3 ;
    double c3 = b3 ;
    double c4 = b4 + b5 ;
    double c5 = b5 + b6 ;
    double c6 = b6 + b7 ;
    double c7 = b7 ;

    double d0  = c0 ;
    double d1  = c1 ;
    double d2  = c2 ;
    double d3  = c3 ;
    double d4  = c4 + c6 ;
    double d5  = c5 ;
    double d6  = c6 ;
    double d7  = c7 ;

    double e0 = d0 + d1 ;
    double e1 = d0 - d1 ;
    double e2 = FC4 * d2 ;
    double e3 = d3 ;
    double e4 = FC4 * d4 ;
    double e5 = FC4 * d5 ;
    double e6 = d6 ;
    double e7 = d7 ;

    double f0 = e0 ;
    double f1 = e1 ;
    double f2 = e2 ;
    double f3 = e3 ;
    double f4 = e4 + e6 ;
    double f5 = e5 ;
    double f6 = e4 - e6 ;
    double f7 = e7 ;

    double g0 = f0 ;
    double g1 = f1 ;
    double g2 = f2 ;
    double g3 = f3 ;
    double g4 = FSEC2 * f4 ;
    double g5 = f7 - f5 ;
    double g6 = FSEC6 * f6 ;
    double g7 = f5 + f7 ;

    double h0 = g0 ;
    double h1 = g1 ;
    double h2 = g2 + g3 ;
    double h3 = g3 - g2 ;
    double h4 = g4 + g7 ;
    double h5 = g5 + g6 ;
    double h6 = g5 - g6 ;
    double h7 = g7 - g4 ;
    double i0 = h0 * qt.float_scaling [row * JPEGSAMPLEWIDTH] ;
    double i1 = h4 * qt.float_scaling [row * JPEGSAMPLEWIDTH+1] ;
    double i2 = h2 * qt.float_scaling [row * JPEGSAMPLEWIDTH+2] ;
    double i3 = h6 * qt.float_scaling [row * JPEGSAMPLEWIDTH+3] ;
    double i4 = h1 * qt.float_scaling [row * JPEGSAMPLEWIDTH+4] ;
    double i5 = h5 * qt.float_scaling [row * JPEGSAMPLEWIDTH+5] ;
    double i6 = h3 * qt.float_scaling [row * JPEGSAMPLEWIDTH+6] ;
    double i7 = h7 * qt.float_scaling [row * JPEGSAMPLEWIDTH+7] ;
    if (i0 >= 0.0)
      output [row * JPEGSAMPLEWIDTH] = (BYTE2) (i0 + 0.5) ;
    else
      output [row * JPEGSAMPLEWIDTH] = (BYTE2) (i0 - 0.5) ;
    if (i1 >= 0.0)
      output [row * JPEGSAMPLEWIDTH + 1] = (BYTE2) (i1 + 0.5) ;
    else
      output [row * JPEGSAMPLEWIDTH + 1] = (BYTE2) (i1 - 0.5) ;
    if (i2 >= 0.0)
      output [row * JPEGSAMPLEWIDTH + 2] = (BYTE2) (i2 + 0.5) ;
    else
      output [row * JPEGSAMPLEWIDTH + 2] = (BYTE2) (i2 - 0.5) ;
    if (i3 >= 0.0)
      output [row * JPEGSAMPLEWIDTH + 3] = (BYTE2) (i3 + 0.5) ;
    else
      output [row * JPEGSAMPLEWIDTH + 3] = (BYTE2) (i3 - 0.5) ;
    if (i4 >= 0.0)
      output [row * JPEGSAMPLEWIDTH + 4] = (BYTE2) (i4 + 0.5) ;
    else
      output [row * JPEGSAMPLEWIDTH + 4] = (BYTE2) (i4 - 0.5) ;
    if (i5 >= 0.0)
      output [row * JPEGSAMPLEWIDTH + 5] = (BYTE2) (i5 + 0.5) ;
    else
      output [row * JPEGSAMPLEWIDTH + 5] = (BYTE2) (i5 - 0.5) ;
    if (i6 >= 0.0)
      output [row * JPEGSAMPLEWIDTH + 6] = (BYTE2) (i6 + 0.5) ;
    else
      output [row * JPEGSAMPLEWIDTH + 6] = (BYTE2) (i6 - 0.5) ;
    if (i7 >= 0.0)
      output [row * JPEGSAMPLEWIDTH + 7] = (BYTE2) (i7 + 0.5) ;
    else
      output [row * JPEGSAMPLEWIDTH + 7] = (BYTE2) (i7 - 0.5) ;
  }
  return ;
}

} // Namespace ColosseumPrivate