home *** CD-ROM | disk | FTP | other *** search
/ Developer CD Series 2000 August: Tool Chest / Dev.CD Aug 00 TC Disk 1.toast / pc / sample code / devices and hardware / velocity engine / veleng wavelet / source / swavelet_int.c < prev    next >
Encoding:
C/C++ Source or Header  |  2000-06-23  |  19.5 KB  |  771 lines
  1. /*
  2.     File:        SWavelet_int.c
  3.  
  4.     Contains:    Scalar integer implementation Daubechies D4 wavelet code
  5.  
  6.     Copyright:    Copyright © 2000 by Apple Computer, Inc., All Rights Reserved.
  7.  
  8.     Disclaimer:    IMPORTANT:  This Apple software is supplied to you by Apple Computer, Inc.
  9.                 ("Apple") in consideration of your agreement to the following terms, and your
  10.                 use, installation, modification or redistribution of this Apple software
  11.                 constitutes acceptance of these terms.  If you do not agree with these terms,
  12.                 please do not use, install, modify or redistribute this Apple software.
  13.  
  14.                 In consideration of your agreement to abide by the following terms, and subject
  15.                 to these terms, Apple grants you a personal, non-exclusive license, under Apple’s
  16.                 copyrights in this original Apple software (the "Apple Software"), to use,
  17.                 reproduce, modify and redistribute the Apple Software, with or without
  18.                 modifications, in source and/or binary forms; provided that if you redistribute
  19.                 the Apple Software in its entirety and without modifications, you must retain
  20.                 this notice and the following text and disclaimers in all such redistributions of
  21.                 the Apple Software.  Neither the name, trademarks, service marks or logos of
  22.                 Apple Computer, Inc. may be used to endorse or promote products derived from the
  23.                 Apple Software without specific prior written permission from Apple.  Except as
  24.                 expressly stated in this notice, no other rights or licenses, express or implied,
  25.                 are granted by Apple herein, including but not limited to any patent rights that
  26.                 may be infringed by your derivative works or by other works in which the Apple
  27.                 Software may be incorporated.
  28.  
  29.                 The Apple Software is provided by Apple on an "AS IS" basis.  APPLE MAKES NO
  30.                 WARRANTIES, EXPRESS OR IMPLIED, INCLUDING WITHOUT LIMITATION THE IMPLIED
  31.                 WARRANTIES OF NON-INFRINGEMENT, MERCHANTABILITY AND FITNESS FOR A PARTICULAR
  32.                 PURPOSE, REGARDING THE APPLE SOFTWARE OR ITS USE AND OPERATION ALONE OR IN
  33.                 COMBINATION WITH YOUR PRODUCTS.
  34.  
  35.                 IN NO EVENT SHALL APPLE BE LIABLE FOR ANY SPECIAL, INDIRECT, INCIDENTAL OR
  36.                 CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE
  37.                 GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  38.                 ARISING IN ANY WAY OUT OF THE USE, REPRODUCTION, MODIFICATION AND/OR DISTRIBUTION
  39.                 OF THE APPLE SOFTWARE, HOWEVER CAUSED AND WHETHER UNDER THEORY OF CONTRACT, TORT
  40.                 (INCLUDING NEGLIGENCE), STRICT LIABILITY OR OTHERWISE, EVEN IF APPLE HAS BEEN
  41.                 ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  42.  
  43. */
  44.  
  45. #ifdef __MWERKS__
  46. #include <altivec.h>
  47. #endif
  48.  
  49. #include <MacTypes.h>
  50. #include <MacMemory.h>
  51.  
  52. #include "sWavelet_int.h"
  53.  
  54.  
  55. #define DO_HORIZ 1
  56. #define DO_VERT 1
  57.  
  58. #define INT_WAVELET_DO_ENDS    1
  59. #define INT_SHIFT_AMOUNT    0
  60.  
  61. #define    H0    (11 * (1<< INT_SHIFT_AMOUNT))
  62. #define    H1    (19 * (1<< INT_SHIFT_AMOUNT))
  63. #define    H2    (5 * (1<< INT_SHIFT_AMOUNT))
  64. #define    H3    (-3 * (1<< INT_SHIFT_AMOUNT))
  65.  
  66. #define INT_WAVELET_SHIFT_AMOUNT        4
  67. #define INVERSE_WAVELET_SHIFT_AMOUNT    5
  68.  
  69. #define INT_WAVELET_DIVIDE_AMOUNT            (1 << INT_WAVELET_SHIFT_AMOUNT)
  70. #define INVERSE_WAVELET_DIVIDE_AMOUNT        (1 << INVERSE_WAVELET_SHIFT_AMOUNT)
  71. #define HALF_SHIFT_ADD                        (1 << (INT_WAVELET_SHIFT_AMOUNT-1))
  72.  
  73.  
  74. static void ExpandQuad8To16(    unsigned char *pSrc,
  75.                                 unsigned short *pDst,
  76.                                 unsigned long length)
  77. {
  78.     int                    vectorInIndex;
  79.     
  80.     for (vectorInIndex = 0; vectorInIndex < length; vectorInIndex++) {
  81.         *pDst++ = *pSrc++;
  82.     }
  83. }                            
  84.  
  85. static void Pack16To8(     signed short     *pSrc,
  86.                     unsigned char *pDest,
  87.                     unsigned long    length)
  88. {
  89.     signed short        element;
  90.     int                    i;
  91.     
  92.     for (i = 0; i<length; i++) {
  93.         element = *pSrc++;
  94.         
  95.         if (element < 0) element = 0;
  96.         if (element > 0xff) element = 0xff;
  97.         
  98.         *pDest++ = element;    
  99.     }
  100. }                        
  101.  
  102. void PadXY_32Bit(unsigned long *pSrc, Handle h2, unsigned long rowBytes, long x, long y, long padX, long padY)
  103. {
  104.     unsigned long     *pInQuad;
  105.     unsigned long    *pOutQuad;
  106.     unsigned long    outQuads = padX*padY;
  107.     unsigned long    inQuads = x*y;
  108.     int                i,j;
  109.     
  110.     ::SetHandleSize(h2, padX*padY*4);
  111.     ThrowIfMemError_();
  112.     
  113.     pInQuad = pSrc;
  114.     pOutQuad = (unsigned long*)(*h2);
  115.  
  116.     for (i=0; i<y; i++) {
  117.         pInQuad = pSrc + ((rowBytes/4)*i);
  118.         
  119.         for (j=0; j<x; j++) {
  120.             *pOutQuad++ = *pInQuad++;
  121.         }        
  122.  
  123.         for (; j<padX; j++) {
  124.             *pOutQuad++ = 0;
  125.         }
  126.     }
  127.     
  128.     for (;i<padY; i++) {
  129.         for (j=0; j<padX; j++) {
  130.             *pOutQuad++ = 0;
  131.         }            
  132.     }
  133. }
  134.  
  135.  
  136. #pragma mark -
  137.  
  138.  
  139. void sFWVT_4_Quad16_Int_Vertical(short *pSrc,
  140.                             short *pDest,
  141.                             unsigned long numQuads,
  142.                             unsigned long skipQuadCount,
  143.                             unsigned long    columns)                
  144. {
  145.     long                    vectorInIndex;
  146.     long                    columnIndex;
  147.     long                    length = (numQuads/2);
  148.     
  149.     signed short            *pSrcInput;
  150.     signed short            *pLoOutput;
  151.     signed short            *pHiOutput;
  152.  
  153.     signed short            A_First;
  154.     signed short            A_Second;
  155.  
  156.     signed short            element0;
  157.     signed short            element1;
  158.     signed short            element2;
  159.     signed short            element3;
  160.  
  161.     signed long                result0;
  162.     signed long                result1;
  163.  
  164.     for (columnIndex = 0; columnIndex < columns*4; columnIndex++) {
  165.  
  166.         pSrcInput = pSrc + (columnIndex);
  167.         pLoOutput = pDest+ (columnIndex);
  168.         pHiOutput = pLoOutput + ((skipQuadCount*4)*(numQuads/2));
  169.  
  170.         A_First = element0 = *pSrcInput;
  171.         pSrcInput += (skipQuadCount*4);
  172.         
  173.         A_Second = element1 = *pSrcInput;
  174.         pSrcInput += (skipQuadCount*4);
  175.  
  176.         element2 = *pSrcInput;
  177.         pSrcInput += (skipQuadCount*4);
  178.  
  179.         element3 = *pSrcInput;
  180.         pSrcInput += (skipQuadCount*4);
  181.  
  182.         result0 = (HALF_SHIFT_ADD) + (element0*H0)+(element1*H1)+(element2*H2)+(element3*H3);
  183.         result1 = (HALF_SHIFT_ADD) + (element0*H3)-(element1*H2)+(element2*H1)-(element3*H0);
  184.         
  185.         result0 /= INT_WAVELET_DIVIDE_AMOUNT;
  186.         result1 /= INT_WAVELET_DIVIDE_AMOUNT;
  187.  
  188.         *pLoOutput = result0;
  189.         *pHiOutput = result1;
  190.                 
  191.         pLoOutput += (skipQuadCount*4);        
  192.         pHiOutput += (skipQuadCount*4);        
  193.         
  194.         element0 = element2;
  195.         element1 = element3;
  196.         
  197.         element2 = *pSrcInput;
  198.         pSrcInput += (skipQuadCount*4);
  199.  
  200.         element3 = *pSrcInput;
  201.         pSrcInput += (skipQuadCount*4);
  202.         
  203.         for (vectorInIndex = 0; vectorInIndex < length-2; vectorInIndex++) {
  204.         
  205.             result0 = (HALF_SHIFT_ADD) + (element0*H0)+(element1*H1)+(element2*H2)+(element3*H3);
  206.             result1 = (HALF_SHIFT_ADD) + (element0*H3)-(element1*H2)+(element2*H1)-(element3*H0);
  207.             
  208.             result0 /= INT_WAVELET_DIVIDE_AMOUNT;
  209.             result1 /= INT_WAVELET_DIVIDE_AMOUNT;
  210.  
  211.             *pLoOutput = result0;
  212.             *pHiOutput = result1;
  213.                     
  214.             pLoOutput += (skipQuadCount*4);        
  215.             pHiOutput += (skipQuadCount*4);        
  216.             
  217.             element0 = element2;
  218.             element1 = element3;
  219.             
  220.             element2 = *pSrcInput;
  221.             pSrcInput += (skipQuadCount*4);
  222.  
  223.             element3 = *pSrcInput;
  224.             pSrcInput += (skipQuadCount*4);
  225.             
  226.         }
  227.                 
  228.         element2 = A_First;
  229.         element3 = A_Second;
  230.         
  231.         result0 = (HALF_SHIFT_ADD) + (element0*H0)+(element1*H1)+(element2*H2)+(element3*H3);
  232.         result1 = (HALF_SHIFT_ADD) + (element0*H3)-(element1*H2)+(element2*H1)-(element3*H0);
  233.         
  234.         result0 /= INT_WAVELET_DIVIDE_AMOUNT;
  235.         result1 /= INT_WAVELET_DIVIDE_AMOUNT;
  236.  
  237.         *pLoOutput = result0;
  238.         *pHiOutput = result1;
  239.     }
  240. }
  241.  
  242.  
  243. void sFWVT_4_Quad16_Int(    signed short *pSrc,
  244.                             signed short *pDest,
  245.                             unsigned long numQuads,
  246.                             unsigned long skipQuads,
  247.                             unsigned long numRows)                
  248. {
  249.     long                    vectorInIndex;
  250.     long                    elementIndex;
  251.     long                    rowIndex;
  252.     
  253.     signed short            *pSrcInput;
  254.     signed short            *pLoOutput;
  255.     signed short            *pHiOutput;
  256.     
  257.     signed short            First_A;
  258.     signed short            First_R;
  259.     signed short            First_G;
  260.     signed short            First_B;
  261.  
  262.     signed short            Second_A;
  263.     signed short            Second_R;
  264.     signed short            Second_G;
  265.     signed short            Second_B;
  266.  
  267.     signed short            element0;
  268.     signed short            element1;
  269.     signed short            element2;
  270.     signed short            element3;
  271.  
  272.     signed long                result0;
  273.     signed long                result1;
  274.     
  275.     for (rowIndex = 0; rowIndex < numRows; rowIndex++) {    
  276.         pSrcInput = pSrc+(rowIndex*(skipQuads*4));
  277.         pLoOutput = pDest+(rowIndex*(skipQuads*4));
  278.         pHiOutput = pLoOutput + (numQuads*2);
  279.  
  280.         ///////////////////////////////
  281.  
  282.         First_A = element0 = *pSrcInput;
  283.         Second_A = element1 = *(pSrcInput+4);
  284.         element2 = *(pSrcInput+8);
  285.         element3 = *(pSrcInput+12);
  286.         
  287.         result0 = (HALF_SHIFT_ADD) + (element0*H0)+(element1*H1)+(element2*H2)+(element3*H3);
  288.         result1 = (HALF_SHIFT_ADD) + (element0*H3)-(element1*H2)+(element2*H1)-(element3*H0);
  289.         
  290.         result0 /= INT_WAVELET_DIVIDE_AMOUNT;
  291.         result1 /= INT_WAVELET_DIVIDE_AMOUNT;
  292.  
  293.         *pLoOutput++ = result0;
  294.         *pHiOutput++ = result1;
  295.  
  296.         pSrcInput++;
  297.         
  298.         ///////////////////////////////
  299.  
  300.         First_R = element0 = *pSrcInput;
  301.         Second_R = element1 = *(pSrcInput+4);
  302.         element2 = *(pSrcInput+8);
  303.         element3 = *(pSrcInput+12);
  304.         
  305.         result0 = (HALF_SHIFT_ADD) + (element0*H0)+(element1*H1)+(element2*H2)+(element3*H3);
  306.         result1 = (HALF_SHIFT_ADD) + (element0*H3)-(element1*H2)+(element2*H1)-(element3*H0);
  307.         
  308.         result0 /= INT_WAVELET_DIVIDE_AMOUNT;
  309.         result1 /= INT_WAVELET_DIVIDE_AMOUNT;
  310.  
  311.         *pLoOutput++ = result0;
  312.         *pHiOutput++ = result1;
  313.  
  314.         pSrcInput++;
  315.  
  316.         ///////////////////////////////
  317.         
  318.         First_G = element0 = *pSrcInput;
  319.         Second_G = element1 = *(pSrcInput+4);
  320.         element2 = *(pSrcInput+8);
  321.         element3 = *(pSrcInput+12);
  322.         
  323.         result0 = (HALF_SHIFT_ADD) + (element0*H0)+(element1*H1)+(element2*H2)+(element3*H3);
  324.         result1 = (HALF_SHIFT_ADD) + (element0*H3)-(element1*H2)+(element2*H1)-(element3*H0);
  325.         
  326.         result0 /= INT_WAVELET_DIVIDE_AMOUNT;
  327.         result1 /= INT_WAVELET_DIVIDE_AMOUNT;
  328.  
  329.         *pLoOutput++ = result0;
  330.         *pHiOutput++ = result1;
  331.         
  332.         pSrcInput++;
  333.  
  334.         ///////////////////////////////
  335.         
  336.         First_B = element0 = *pSrcInput;
  337.         Second_B = element1 = *(pSrcInput+4);
  338.         element2 = *(pSrcInput+8);
  339.         element3 = *(pSrcInput+12);
  340.         
  341.         result0 = (HALF_SHIFT_ADD) + (element0*H0)+(element1*H1)+(element2*H2)+(element3*H3);
  342.         result1 = (HALF_SHIFT_ADD) + (element0*H3)-(element1*H2)+(element2*H1)-(element3*H0);
  343.         
  344.         result0 /= INT_WAVELET_DIVIDE_AMOUNT;
  345.         result1 /= INT_WAVELET_DIVIDE_AMOUNT;
  346.  
  347.         *pLoOutput++ = result0;
  348.         *pHiOutput++ = result1;
  349.         
  350.         pSrcInput += 5;
  351.         
  352.         for (vectorInIndex = 0; vectorInIndex < ((numQuads-4)/2); vectorInIndex++) {
  353.             for (elementIndex = 0; elementIndex < 4; elementIndex++) {
  354.  
  355.                 element0 = *pSrcInput;
  356.                 element1 = *(pSrcInput+4);
  357.                 element2 = *(pSrcInput+8);
  358.                 element3 = *(pSrcInput+12);
  359.                 
  360.                 result0 = (HALF_SHIFT_ADD) + (element0*H0)+(element1*H1)+(element2*H2)+(element3*H3);
  361.                 result1 = (HALF_SHIFT_ADD) + (element0*H3)-(element1*H2)+(element2*H1)-(element3*H0);
  362.                 
  363.                 result0 /= INT_WAVELET_DIVIDE_AMOUNT;
  364.                 result1 /= INT_WAVELET_DIVIDE_AMOUNT;                
  365.  
  366.                 *pLoOutput++ = result0;
  367.                 *pHiOutput++ = result1;
  368.                 
  369.                 pSrcInput++;
  370.             }
  371.             
  372.             pSrcInput += 4;
  373.             
  374.         }
  375.         
  376.         
  377.         ///////////////////////////////
  378.  
  379.         element0 = *pSrcInput;
  380.         element1 = *(pSrcInput+4);
  381.         element2 = First_A;
  382.         element3 = Second_A;
  383.         
  384.         result0 = (HALF_SHIFT_ADD) + (element0*H0)+(element1*H1)+(element2*H2)+(element3*H3);
  385.         result1 = (HALF_SHIFT_ADD) + (element0*H3)-(element1*H2)+(element2*H1)-(element3*H0);
  386.         
  387.         result0 /= INT_WAVELET_DIVIDE_AMOUNT;
  388.         result1 /= INT_WAVELET_DIVIDE_AMOUNT;
  389.  
  390.         *pLoOutput++ = result0;
  391.         *pHiOutput++ = result1;
  392.  
  393.         pSrcInput++;
  394.         
  395.         ///////////////////////////////
  396.  
  397.         element0 = *pSrcInput;
  398.         element1 = *(pSrcInput+4);
  399.         element2 = First_R;
  400.         element3 = Second_R;
  401.         
  402.         result0 = (HALF_SHIFT_ADD) + (element0*H0)+(element1*H1)+(element2*H2)+(element3*H3);
  403.         result1 = (HALF_SHIFT_ADD) + (element0*H3)-(element1*H2)+(element2*H1)-(element3*H0);
  404.         
  405.         result0 /= INT_WAVELET_DIVIDE_AMOUNT;
  406.         result1 /= INT_WAVELET_DIVIDE_AMOUNT;
  407.  
  408.         *pLoOutput++ = result0;
  409.         *pHiOutput++ = result1;
  410.  
  411.         pSrcInput++;
  412.         
  413.         ///////////////////////////////
  414.  
  415.         element0 = *pSrcInput;
  416.         element1 = *(pSrcInput+4);
  417.         element2 = First_G;
  418.         element3 = Second_G;
  419.         
  420.         result0 = (HALF_SHIFT_ADD) + (element0*H0)+(element1*H1)+(element2*H2)+(element3*H3);
  421.         result1 = (HALF_SHIFT_ADD) + (element0*H3)-(element1*H2)+(element2*H1)-(element3*H0);
  422.         
  423.         result0 /= INT_WAVELET_DIVIDE_AMOUNT;
  424.         result1 /= INT_WAVELET_DIVIDE_AMOUNT;
  425.  
  426.         *pLoOutput++ = result0;
  427.         *pHiOutput++ = result1;
  428.  
  429.         pSrcInput++;
  430.  
  431.         ///////////////////////////////
  432.  
  433.         element0 = *pSrcInput;
  434.         element1 = *(pSrcInput+4);
  435.         element2 = First_B;
  436.         element3 = Second_B;
  437.         
  438.         result0 = (HALF_SHIFT_ADD) + (element0*H0)+(element1*H1)+(element2*H2)+(element3*H3);
  439.         result1 = (HALF_SHIFT_ADD) + (element0*H3)-(element1*H2)+(element2*H1)-(element3*H0);
  440.         
  441.         result0 /= INT_WAVELET_DIVIDE_AMOUNT;
  442.         result1 /= INT_WAVELET_DIVIDE_AMOUNT;
  443.  
  444.         *pLoOutput = result0;
  445.         *pHiOutput = result1;        
  446.     }    
  447. }
  448.  
  449. void sIFWVT_4_Quad16_Int_Vertical(    signed short *pSrc,
  450.                             signed short *pDest,
  451.                             unsigned long numQuads,
  452.                             unsigned long skipQuadCount,
  453.                             unsigned long columns)                
  454. {
  455.     long                    vectorInIndex;
  456.     long                    columnIndex;
  457.     long                    length = (numQuads/2);
  458.     long                    skipShorts = skipQuadCount * 4;
  459.     
  460.     signed short            *pSrcLo;
  461.     signed short            *pSrcHi;
  462.     signed short            *pColumnOut;
  463.     
  464.     signed short            element0;
  465.     signed short            element1;
  466.     signed short            element2;
  467.     signed short            element3;
  468.     
  469.     signed long                result0;
  470.     signed long                result1;
  471.     
  472.     for (columnIndex = 0; columnIndex < columns*4; columnIndex++) {
  473.  
  474.         // set up pointers: 
  475.         pColumnOut     = pDest + columnIndex;
  476.         pSrcLo         = pSrc    + columnIndex;
  477.         pSrcHi         = pSrcLo + ((skipShorts)*(numQuads/2));
  478.                 
  479.         element0    = *pSrcLo;
  480.         element1    = *(pSrcHi - skipShorts);
  481.         element2     = *pSrcHi;
  482.         element3    = *(pSrcLo+((numQuads-1)*(skipShorts)));
  483.         
  484.         pSrcHi        +=skipShorts;
  485.         pSrcLo        +=skipShorts;
  486.  
  487.         result0 = (HALF_SHIFT_ADD) + (element0*H0)+(element1*H2)+(element2*H3)+(element3*H1);
  488.         result1 = (HALF_SHIFT_ADD) + (element0*H1)+(element1*H3)-(element2*H2)-(element3*H0);
  489.  
  490.         result0 /= INVERSE_WAVELET_DIVIDE_AMOUNT;
  491.         result1 /= INVERSE_WAVELET_DIVIDE_AMOUNT;
  492.  
  493.         *pColumnOut = result0;
  494.         pColumnOut += skipShorts;
  495.  
  496.         *pColumnOut = result1;
  497.         pColumnOut += skipShorts;
  498.             
  499.         for (vectorInIndex = 0; vectorInIndex < length-1; vectorInIndex++) {
  500.             element1    = *pSrcLo;
  501.             element3     = *pSrcHi;
  502.  
  503.             pSrcHi        +=skipShorts;
  504.             pSrcLo        +=skipShorts;
  505.  
  506.             result0 = (HALF_SHIFT_ADD) + (element0*H2)+(element1*H0)+(element2*H1)+(element3*H3);
  507.             result1 = (HALF_SHIFT_ADD) + (element0*H3)+(element1*H1)-(element2*H0)-(element3*H2);
  508.  
  509.             result0 /= INVERSE_WAVELET_DIVIDE_AMOUNT;
  510.             result1 /= INVERSE_WAVELET_DIVIDE_AMOUNT;
  511.  
  512.             *pColumnOut = result0;
  513.             pColumnOut += skipShorts;
  514.  
  515.             *pColumnOut = result1;
  516.             pColumnOut += skipShorts;
  517.  
  518.             element0 = element1;
  519.             element2 = element3;
  520.             
  521.         }
  522.  
  523.     }    
  524. }
  525.  
  526.  
  527. void sIFWVT_4_Quad16_Int(    signed short *pSrc,
  528.                             signed short *pDest,
  529.                             unsigned long numQuads,
  530.                             unsigned long skipQuads,
  531.                             unsigned long numRows)                
  532. {
  533.     long                         length = (numQuads/2);
  534.     long                        vectorIndex;
  535.     long                        rowIndex;
  536.     long                        skipShorts = skipQuads*4;
  537.     long                        elementIndex;
  538.     
  539.     signed short                *pRowDest;
  540.     signed short                *pSrcHi, *pSrcLo;
  541.  
  542.     signed short                element0;
  543.     signed short                element1;
  544.     signed short                element2;
  545.     signed short                element3;
  546.  
  547.     signed long                    result0;
  548.     signed long                    result1;
  549.  
  550.     for (rowIndex = 0; rowIndex < numRows; rowIndex++) {
  551.         
  552.         // set up pointers:
  553.         pSrcLo = pSrc + (rowIndex*skipShorts);
  554.         pSrcHi = pSrcLo + (numQuads/2)*4;
  555.         pRowDest = pDest + (rowIndex*skipShorts);
  556.  
  557.         ////////////////////////////////////
  558.                 
  559.         element0    = *pSrcLo;
  560.         element1    = *(pSrcHi - 4);
  561.         element2     = *pSrcHi;
  562.         element3    = *(pSrcLo+((numQuads-1)*4));
  563.         
  564.         result0 = (HALF_SHIFT_ADD) + (element0*H0)+(element1*H2)+(element2*H3)+(element3*H1);
  565.         result1 = (HALF_SHIFT_ADD) + (element0*H1)+(element1*H3)-(element2*H2)-(element3*H0);
  566.  
  567.         result0 /= INVERSE_WAVELET_DIVIDE_AMOUNT;
  568.         result1 /= INVERSE_WAVELET_DIVIDE_AMOUNT;
  569.         
  570.         *pRowDest    = result0;
  571.         *((pRowDest++)+4)    = result1;
  572.  
  573.         ////////////////////////////////////
  574.                 
  575.         element0    = *(pSrcLo+1);
  576.         element1    = *(pSrcHi - 3);
  577.         element2     = *(pSrcHi+1);
  578.         element3    = *(pSrcLo+((numQuads-1)*4)+1);
  579.         
  580.         result0 = (HALF_SHIFT_ADD) + (element0*H0)+(element1*H2)+(element2*H3)+(element3*H1);
  581.         result1 = (HALF_SHIFT_ADD) + (element0*H1)+(element1*H3)-(element2*H2)-(element3*H0);
  582.  
  583.         result0 /= INVERSE_WAVELET_DIVIDE_AMOUNT;
  584.         result1 /= INVERSE_WAVELET_DIVIDE_AMOUNT;
  585.         
  586.         *pRowDest    = result0;
  587.         *((pRowDest++)+4)    = result1;
  588.  
  589.         ////////////////////////////////////
  590.                 
  591.         element0    = *(pSrcLo+2);
  592.         element1    = *(pSrcHi - 2);
  593.         element2     = *(pSrcHi+2);
  594.         element3    = *(pSrcLo+((numQuads-1)*4)+2);
  595.         
  596.         result0 = (HALF_SHIFT_ADD) + (element0*H0)+(element1*H2)+(element2*H3)+(element3*H1);
  597.         result1 = (HALF_SHIFT_ADD) + (element0*H1)+(element1*H3)-(element2*H2)-(element3*H0);
  598.  
  599.         result0 /= INVERSE_WAVELET_DIVIDE_AMOUNT;
  600.         result1 /= INVERSE_WAVELET_DIVIDE_AMOUNT;
  601.         
  602.         *pRowDest    = result0;
  603.         *((pRowDest++)+4)    = result1;
  604.  
  605.         ////////////////////////////////////
  606.                 
  607.         element0    = *(pSrcLo+3);
  608.         element1    = *(pSrcHi - 1);
  609.         element2     = *(pSrcHi+3);
  610.         element3    = *(pSrcLo+((numQuads-1)*4)+3);
  611.         
  612.         result0 = (HALF_SHIFT_ADD) + (element0*H0)+(element1*H2)+(element2*H3)+(element3*H1);
  613.         result1 = (HALF_SHIFT_ADD) + (element0*H1)+(element1*H3)-(element2*H2)-(element3*H0);
  614.  
  615.         result0 /= INVERSE_WAVELET_DIVIDE_AMOUNT;
  616.         result1 /= INVERSE_WAVELET_DIVIDE_AMOUNT;
  617.         
  618.         *pRowDest    = result0;
  619.         *((pRowDest++)+4)    = result1;
  620.  
  621.         pRowDest += 4;
  622.         
  623.         for (vectorIndex = 0; vectorIndex < (numQuads-2)/2; vectorIndex++) {
  624.             for (elementIndex = 0; elementIndex < 4; elementIndex++) {
  625.                 element0 = *pSrcLo;
  626.                 element1 = *(pSrcLo + 4);
  627.                 element2 = *pSrcHi;            
  628.                 element3 = *(pSrcHi + 4);            
  629.                 
  630.                 result0 = (HALF_SHIFT_ADD) + (element0*H2)+(element1*H0)+(element2*H1)+(element3*H3);
  631.                 result1 = (HALF_SHIFT_ADD) + (element0*H3)+(element1*H1)-(element2*H0)-(element3*H2);
  632.  
  633.                 result0 /= INVERSE_WAVELET_DIVIDE_AMOUNT;
  634.                 result1 /= INVERSE_WAVELET_DIVIDE_AMOUNT;
  635.                 
  636.                 *pRowDest            = result0;
  637.                 *((pRowDest++)+4)    = result1;
  638.  
  639.                 pSrcLo++;
  640.                 pSrcHi++;
  641.             }
  642.  
  643.             pRowDest += 4;
  644.             
  645.         }
  646.  
  647.     }
  648.  
  649. }
  650.  
  651.  
  652.  
  653. void sFWVT_4_Quad16_2DInt(signed short *pSrc,
  654.                         signed short *pDst,
  655.                         signed short *pTemp,
  656.                         unsigned long x,
  657.                         unsigned long y,
  658.                         unsigned long rowQuads,
  659.                         unsigned long depth)
  660. {
  661.     unsigned long     depthIndex;
  662.     unsigned long    currentDepthHeight;
  663.     unsigned long    currentDepthWidth;
  664.         
  665.  
  666.     currentDepthHeight = y;
  667.     currentDepthWidth = x;
  668.  
  669.  
  670.     ExpandQuad8To16((unsigned char*)pSrc, (unsigned short*)pDst, y*rowQuads*4);
  671.  
  672.     sFWVT_4_Quad16_Int(pDst,
  673.                     pTemp,
  674.                     currentDepthWidth,
  675.                     rowQuads,
  676.                     currentDepthHeight
  677.                     );                
  678.  
  679.     sFWVT_4_Quad16_Int_Vertical(pTemp,
  680.                     pDst,
  681.                     currentDepthHeight,
  682.                     rowQuads,
  683.                     currentDepthWidth);                
  684.  
  685.  
  686.     for (depthIndex = 1; depthIndex < depth; depthIndex++) {
  687.         currentDepthHeight >>= 1;        
  688.         currentDepthWidth >>= 1;        
  689.     
  690.  
  691.         sFWVT_4_Quad16_Int(    pDst,
  692.                         pTemp,
  693.                         currentDepthWidth,
  694.                         rowQuads,
  695.                         currentDepthHeight
  696.                         );                
  697.  
  698.         sFWVT_4_Quad16_Int_Vertical(pTemp,
  699.                         pDst,
  700.                         currentDepthHeight,
  701.                         rowQuads,
  702.                         currentDepthWidth);                
  703.  
  704.  
  705.     
  706.     }        
  707. }
  708.  
  709.  
  710.  
  711. void sIFWVT_4_Quad16_2DInt(signed short *pSrc,
  712.                         signed short *pDst,
  713.                         signed short *pTemp,
  714.                         unsigned long x,
  715.                         unsigned long y,
  716.                         unsigned long rowQuads,
  717.                         unsigned long depth)                
  718. {
  719.     unsigned long     depthIndex;
  720.     unsigned long    currentDepthHeight;
  721.     unsigned long    currentDepthWidth;
  722.     
  723.     
  724.     currentDepthHeight = y >> (depth);
  725.     currentDepthWidth =  x >> (depth);
  726.  
  727.     if (depth) {
  728.  
  729.         for (depthIndex = 0; depthIndex < depth-1; depthIndex++) {
  730.             currentDepthHeight     <<= 1;
  731.             currentDepthWidth     <<= 1;
  732.             
  733.         sIFWVT_4_Quad16_Int_Vertical(pSrc,
  734.                         pTemp,
  735.                         currentDepthHeight,
  736.                         rowQuads,
  737.                         currentDepthWidth);            
  738.  
  739.         sIFWVT_4_Quad16_Int(pTemp,
  740.                         pSrc,
  741.                         currentDepthWidth,
  742.                         rowQuads,
  743.                         currentDepthHeight
  744.                         );            
  745.                 
  746.         }
  747.         
  748.         currentDepthHeight     <<= 1;
  749.         currentDepthWidth     <<= 1;
  750.  
  751.         sIFWVT_4_Quad16_Int_Vertical(pSrc,
  752.                         pTemp,
  753.                         currentDepthHeight,
  754.                         rowQuads,
  755.                         currentDepthWidth);            
  756.  
  757.         sIFWVT_4_Quad16_Int(pTemp,
  758.                         pSrc,
  759.                         currentDepthWidth,
  760.                         rowQuads,
  761.                         currentDepthHeight
  762.                         );    
  763.                         
  764.         Pack16To8(pSrc, (unsigned char *)pDst, y * rowQuads * 4);        
  765.     
  766.     }
  767.  
  768. }
  769.  
  770.  
  771.