vopseenc.cpp

网络MPEG4IP流媒体开发源代码

/*************************************************************************

This software module was originally developed by 

	Ming-Chieh Lee ([e-mail]), Microsoft Corporation
	Wei-ge Chen ([e-mail]), Microsoft Corporation
	Bruce Lin ([e-mail]), Microsoft Corporation
	Chuang Gu ([e-mail]), Microsoft Corporation
	Simon Winder ([e-mail]), Microsoft Corporation
	(date: March, 1996)
and edited by
        Wei Wu ([e-mail]) Rockwell Science Center

and also edited by
	Yoshihiro Kikuchi (TOSHIBA CORPORATION)
	Takeshi Nagai (TOSHIBA CORPORATION)
	Toshiaki Watanabe (TOSHIBA CORPORATION)
	Noboru Yamaguchi (TOSHIBA CORPORATION)

and also edited by
	David B. Shu ([e-mail]), Hughes Electronics/HRL Laboratories

	Marc Mongenet ([e-mail]), Swiss Federal Institute of Technology, Lausanne (EPFL)

and also edited by
	Yoshinori Suzuki (Hitachi, Ltd.)

in the course of development of the MPEG-4 Video (ISO/IEC 14496-2). 
This software module is an implementation of a part of one or more MPEG-4 Video tools 
as specified by the MPEG-4 Video. 
ISO/IEC gives users of the MPEG-4 Video free license to this software module or modifications 
thereof for use in hardware or software products claiming conformance to the MPEG-4 Video. 
Those intending to use this software module in hardware or software products are advised that its use may infringe existing patents. 
The original developer of this software module and his/her company, 
the subsequent editors and their companies, 
and ISO/IEC have no liability for use of this software module or modifications thereof in an implementation. 
Copyright is not released for non MPEG-4 Video conforming products. 
Microsoft retains full right to use the code for his/her own purpose, 
assign or donate the code to a third party and to inhibit third parties from using the code for non <MPEG standard> conforming products. 
This copyright notice must be included in all copies or derivative works. 

Copyright (c) 1996, 1997.

Module Name:

	vopSeEnc.hpp

Abstract:

	Encoder for one VO.

Revision History:
	Sept. 30, 1997: Error resilient tools added by Toshiba
	Dec.  11, 1997:	Interlaced tools added by NextLevel Systems
					X. Chen, [e-mail] B. Eifrig, [e-mail]
	Jun.  16, 1998: add Complexity Estimation syntax support
					Marc Mongenet ([e-mail]) - EPFL
	Jan.  28, 1999: Rounding control parameters added by Hitachi, Ltd.
	May    9, 1999:	tm5 rate control by DemoGraFX, [e-mail]

*************************************************************************/

#include <stdio.h>
#include <iostream.h>
#include <fstream.h>
#include <math.h>
#include <stdlib.h>

#include "typeapi.h"
#include "codehead.h"
#include "global.hpp"
#include "entropy/bitstrm.hpp"
#include "entropy/entropy.hpp"
#include "entropy/huffman.hpp"
#include "mode.hpp"
#include "dct.hpp"
#include "tps_enhcbuf.hpp" // added by Sharp (98/2/12)
#include "vopses.hpp"
#include "vopseenc.hpp"
#include "enhcbufenc.hpp" // added by Sharp (98/2/12)
#include "cae.h" // Added for error resilient mode by Toshiba(1997-11-14)

#ifdef __MFC_
#ifdef _DEBUG
#undef THIS_FILE
static char BASED_CODE THIS_FILE[] = __FILE__;
#endif

#define new DEBUG_NEW				   
#endif // __MFC_

CVideoObjectEncoder::~CVideoObjectEncoder ()
{
	delete m_pvopcOrig;
	delete m_pvopcRefOrig1;
	delete m_pvopcRefOrig0;

	delete m_puciRefQZoom0;
	delete m_puciRefQZoom1;
	delete m_pfdct;
	delete [] m_rgdSNR;

	// bitstream stuff
	delete [] m_pchBitsBuffer; 
	delete m_pbitstrmOut;
	delete m_pentrencSet;
	delete [] m_pchShapeBitsBuffer;
	delete m_pbitstrmShape;

//	Added for data partitioning mode By Toshiba(1998-1-16:DP+RVLC)
	delete [] *m_pchShapeBitsBuffer_DP;
	delete [] *m_pbitstrmShape_DP;
	delete m_pchShapeBitsBuffer_DP;
	delete m_pbitstrmShape_DP;
//	End Toshiba(1998-1-16)

	// shape
	delete [] m_rgiSubBlkIndx16x16;
	delete [] m_rgiSSubBlkIndx16x16;
	delete [] m_rgiSubBlkIndx18x18;
	delete [] m_rgiSubBlkIndx20x20;
	delete [] m_rgiPxlIndx12x12;
	delete [] m_rgiPxlIndx8x8;

	// B-VOP MB buffer
	delete m_puciDirectPredMB;
	delete m_puciInterpPredMB;
	delete m_piiDirectErrorMB; 
	delete m_piiInterpErrorMB; 
}

CVideoObjectEncoder::CVideoObjectEncoder (
	UInt uiVOId, 
	VOLMode& volmd, 
	VOPMode& vopmd, 
	UInt nFirstFrame,
	UInt nLastFrame,
	Int iSessionWidth,
	Int iSessionHeight,
	UInt uiRateControl,
	UInt uiBudget,
	ostream* pstrmTrace, // trace outstream
	UInt uiWarpingAccuracy, // for sprite warping
	Int iNumOfPnts, // for sprite warping
	CSiteD** rgstDest, // for sprite warping destination
	SptMode SpriteMode,					// sprite reconstruction mode
	CRct rctFrame,						// sprite warping source
    CRct rctSpt,     	// rct sprite
	Int iMVFileUsage,
	Char* pchMVFileName
) :
	CVideoObject (),
	m_nFirstFrame (nFirstFrame), m_nLastFrame (nLastFrame), 
	m_iBufferSize (uiBudget), m_uiRateControl (uiRateControl),
	m_pvopcOrig (NULL),
	m_rgiSubBlkIndx16x16 (NULL),
	m_rgiSSubBlkIndx16x16 (NULL),
	m_rgiSubBlkIndx18x18 (NULL),
	m_rgiSubBlkIndx20x20 (NULL),
	m_rgiPxlIndx12x12 (NULL),
	m_rgiPxlIndx8x8 (NULL)
{
	m_ivolWidth = iSessionWidth;
	m_ivolHeight = iSessionHeight;		
		
	// sprite stuff
	if (iNumOfPnts >= 0) {
		m_uiSprite = 1;
		m_uiWarpingAccuracy = uiWarpingAccuracy;
		m_iNumOfPnts = iNumOfPnts;
		m_rgstSrcQ = new CSiteD [m_iNumOfPnts];
		m_rgstDstQ = new CSiteD [m_iNumOfPnts];
		m_rctSpt = rctSpt;
		//#ifdef __LOW_LATENCY_SPRITE_
		m_pprgstDest = rgstDest;
		m_sptMode = SpriteMode ;
		m_rctOrg = 	rctFrame;
		m_rctDisplayWindow = m_rctOrg;	//only used for sprite, will be combined with Org later
	}
	else 
		m_uiSprite = 0;

	m_pchBitsBuffer = new Char [iSessionWidth * iSessionHeight * 2];	//we think this is enof for 4:2:0; if crushes, increase the buffer
	m_pbitstrmOut = new COutBitStream (m_pchBitsBuffer, 0, pstrmTrace);
	m_pentrencSet = new CEntropyEncoderSet (*m_pbitstrmOut);

	// shape cache
	m_pchShapeBitsBuffer = new Char [MB_SIZE * MB_SIZE * 2]; // same as above
	m_pbitstrmShape = new COutBitStream (m_pchShapeBitsBuffer, 0, pstrmTrace);
	m_pbitstrmShapeMBOut = m_pbitstrmOut; // initially the output stream (only changes for inter)

	m_uiVOId = uiVOId;
	m_volmd = volmd;

	// NBIT: set right clip table
	setClipTab();

	Int iClockRate = m_volmd.iClockRate;
	assert (iClockRate >= 1 && iClockRate < 65536);
	for (m_iNumBitsTimeIncr = 1; m_iNumBitsTimeIncr < NUMBITS_TIME_RESOLUTION; m_iNumBitsTimeIncr++)	{	
		if (iClockRate == 1)			
			break;
		iClockRate = (iClockRate >> 1);
	}

	m_vopmd = vopmd;
//	decideMVInfo ();                // MV search radius now varys with GOP
	m_vopmd.iRoundingControlEncSwitch = m_volmd.iInitialRoundingType;
	if(!m_volmd.bRoundingControlDisable)
		m_vopmd.iRoundingControlEncSwitch ^= 0x00000001;
	m_vopmd.iRoundingControl = m_vopmd.iRoundingControlEncSwitch;
	Int iMod = iSessionWidth % MB_SIZE;
	Int iSessionWidthRound = (iMod > 0) ? iSessionWidth + MB_SIZE - iMod : iSessionWidth;
	iMod = iSessionHeight % MB_SIZE;
	Int iSessionHeightRound = (iMod > 0) ? iSessionHeight + MB_SIZE - iMod : iSessionHeight;
//	Added for data partitioning mode By Toshiba(1998-1-16:DP+RVLC)
	Int iMB;
	Int iMBN = (iSessionWidthRound/MB_SIZE)*(iSessionHeightRound/MB_SIZE);
	m_pchShapeBitsBuffer_DP = new Char* [iMBN];
	m_pbitstrmShape_DP = new COutBitStream* [iMBN];
	for (iMB = 0; iMB < iMBN; iMB++) {
		m_pchShapeBitsBuffer_DP[iMB] = new Char [MB_SIZE * MB_SIZE * 2]; // same as above
		m_pbitstrmShape_DP[iMB] = new COutBitStream (m_pchShapeBitsBuffer_DP[iMB], 0, pstrmTrace);
	}
// End Toshiba(1998-1-16:DP+RVLC)
	m_rctRefFrameY = CRct (
		-EXPANDY_REF_FRAME, -EXPANDY_REF_FRAME, 
		EXPANDY_REF_FRAME + iSessionWidthRound, EXPANDY_REF_FRAME + iSessionHeightRound
	);
	m_rctRefFrameUV = m_rctRefFrameY.downSampleBy2 ();
	m_pvopcOrig = new CVOPU8YUVBA (m_volmd.fAUsage, m_rctRefFrameY);
	m_pvopcRefOrig0 = new CVOPU8YUVBA (m_volmd.fAUsage, m_rctRefFrameY);
	m_pvopcRefOrig1 = new CVOPU8YUVBA (m_volmd.fAUsage, m_rctRefFrameY);
	allocateVOLMembers (iSessionWidthRound, iSessionHeightRound);

    //#ifdef __LOW_LATENCY_SPRITE_
	// identify initial sprite piece
	if ((m_uiSprite == 1) && (m_sptMode != BASIC_SPRITE) ){
		initialSpritePiece (iSessionWidthRound, iSessionHeightRound);
	}

    CRct rctFrameZoom = m_rctRefFrameY.upSampleBy2 ();
	m_puciRefQZoom0 = new CU8Image (rctFrameZoom);

	m_puciRefQZoom1 = new CU8Image (rctFrameZoom);

	m_pfdct = new CFwdBlockDCT(volmd.nBits);

	// B-VOP MB buffer
	m_puciDirectPredMB = new CU8Image (CRct (0, 0, MB_SIZE, MB_SIZE));
	m_puciInterpPredMB = new CU8Image (CRct (0, 0, MB_SIZE, MB_SIZE));
	m_ppxlcDirectPredMBY = (PixelC*) m_puciDirectPredMB->pixels (); 
	m_ppxlcInterpPredMBY = (PixelC*) m_puciInterpPredMB->pixels (); 

	m_piiDirectErrorMB = new CIntImage (CRct (0, 0, MB_SIZE, MB_SIZE)); 
	m_piiInterpErrorMB = new CIntImage (CRct (0, 0, MB_SIZE, MB_SIZE)); 
	m_ppxliDirectErrorMBY = (PixelI*) m_piiDirectErrorMB->pixels (); 
	m_ppxliInterpErrorMBY = (PixelI*) m_piiInterpErrorMB->pixels (); 

	// with shape
	if (m_volmd.fAUsage != RECTANGLE) {
		m_rgiSubBlkIndx16x16 = computeShapeSubBlkIndex (4, 16);
		m_rgiSSubBlkIndx16x16 = computeShapeSubBlkIndex (2, 16);
		m_rgiSubBlkIndx18x18 = computeShapeSubBlkIndex (4, 18);
		m_rgiSubBlkIndx20x20 = computeShapeSubBlkIndex (4, 20);
		m_rgiPxlIndx12x12 = new Int [8 * 8];		
		Int* piPxl = m_rgiPxlIndx12x12;
		UInt i, j;
		for (i = 2; i < 10; i++)
			for (j = 2; j < 10; j++)
				*piPxl++ = i * 12 + j;
		m_rgiPxlIndx8x8 = new Int [4 * 4];		
		piPxl = m_rgiPxlIndx8x8;
		for (i = 2; i < 6; i++)
			for (j = 2; j < 6; j++)
				*piPxl++ = i * 8 + j;
	}

	m_statsVOL.reset ();
	codeVOHead ();
	codeVOLHead (iSessionWidth, iSessionHeight);//, rctSprite);

	// Added for error resilient mode by Toshiba(1997-11-14): Moved (1998-1-16)
	g_iMaxHeading = MAXHEADING_ERR;
	g_iMaxMiddle = MAXMIDDLE_ERR;
	g_iMaxTrailing = MAXTRAILING_ERR;

	// End Toshiba(1997-11-14)

	m_statsVOL.nBitsStuffing += m_pbitstrmOut->flush ();
	m_statRC.resetSkipMode ();
	m_rgdSNR = (m_volmd.fAUsage == EIGHT_BIT) ? new Double [4] : new Double [3];

	// some fixed variables	
	m_iFrameWidthZoomY = m_iFrameWidthY * 2;
	m_iFrameWidthZoomUV = m_iFrameWidthUV * 2;
	m_iFrameWidthZoomYx2Minus2MB = m_iFrameWidthY * 2 * 2 - 2 * MB_SIZE;
	m_iFrameWidthZoomYx2Minus2Blk = m_iFrameWidthY * 2 * 2 - 2 * BLOCK_SIZE;

	if (m_volmd.fAUsage == RECTANGLE) {
		m_rctCurrVOPY = CRct (0, 0, iSessionWidthRound, iSessionHeightRound);
		m_pvopcOrig->setBoundRct (m_rctCurrVOPY);
		m_rctCurrVOPUV = m_rctCurrVOPY.downSampleBy2 ();
		
		m_rctRefVOPY0 = m_rctCurrVOPY;
		m_rctRefVOPY0.expand (EXPANDY_REFVOP);
		m_rctRefVOPUV0 = m_rctRefVOPY0.downSampleBy2 ();
		
		m_rctRefVOPY1 = m_rctRefVOPY0;
		m_rctRefVOPUV1 = m_rctRefVOPUV0;
		
		m_rctRefVOPZoom0 = m_rctRefVOPY0.upSampleBy2 ();
		m_rctRefVOPZoom1 = m_rctRefVOPY1.upSampleBy2 ();
		m_pvopcRefOrig0->setBoundRct (m_rctRefVOPY0);
		m_pvopcRefOrig1->setBoundRct (m_rctRefVOPY1);

// dshu: begin of modification
	if ((m_uiSprite == 0) || ((m_uiSprite == 1) && (m_sptMode == BASIC_SPRITE)) )	
// dshu: end of modification

		computeVOLConstMembers (); // these VOP members are the same for all frames
	}

	m_vopmd.SpriteXmitMode = STOP ;	
    // tentative solution for indicating the first Sprite VOP
    tentativeFirstSpriteVop = 0;

	// Open motion vector file if used
	m_pchMVFileName = pchMVFileName;
	m_iMVFileUsage = iMVFileUsage;
	if (m_iMVFileUsage != 0) { // wmay - changed from =
		m_fMVFile = fopen(m_pchMVFileName, (m_iMVFileUsage == 1) ? "r" : "w");
		assert(m_fMVFile != NULL);
		m_iMVLineNo = 0;
	}
}

// for back/forward shape
CVideoObjectEncoder::CVideoObjectEncoder (
	UInt uiVOId, 
	VOLMode& volmd, 
	VOPMode& vopmd,
	Int iSessionWidth,
	Int iSessionHeight