sesenc.cpp

此源码是在VC平台下,实现MPEG4编解码的源码

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

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
	(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)

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

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:

	sesEnc.cpp

Abstract:

	Encoder for one video session.

Revision History:
	Sept. 30, 1997: Error resilient tools added by Toshiba
	Nov.  27, 1997: Spatial Scalable tools added, and modified for Spatial Scalable
						by Takefumi Nagumo([e-mail]) SONY
	Dec 11, 1997:	Interlaced tools added by NextLevel Systems (GI)
						(B.Eifrig, [e-mail], X.Chen, [e-mail])
	Jun 17, 1998:	add Complexity Estimation syntax support
					Marc Mongenet (Marc.Mongenet@epfl) - EPFL
	May 9, 1999:	tm5 rate control by DemoGraFX, [e-mail]

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

#include "stdlib.h"
#include "stdio.h"
#include "math.h"
#include "fstream.h"
#include "iostream.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 "sesenc.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)

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

#define new DEBUG_NEW				   
#endif // __MFC_

#ifdef __PC_COMPILER_					//OS specific stuff
#define SUB_CMPFILE "%s\\%2.2d\\%s.cmp"
#define SUB_TRCFILE "%s\\%2.2d\\%s.trc"
#define SUB_YUVFILE	"%s\\%2.2d\\%s.yuv"
#define SUB_SEGFILE	"%s\\%2.2d\\%s.seg"
#define ENHN_SUB_CMPFILE "%s\\%2.2d\\%s_e.cmp"
#define ENHN_SUB_TRCFILE "%s\\%2.2d\\%s_e.trc"
#define ENHN_SUB_YUVFILE "%s\\%2.2d\\%s_e.yuv"
#define ENHN_SUB_SEGFILE "%s\\%2.2d\\%s_e.seg"
#define SUB_VDLFILE "%s\\%2.2d\\%s.spt"	//for sprite i/o
#define SUB_PNTFILE "%s\\%2.2d\\%s.pnt"	//for sprite i/o
#define ROOT_YUVFILE "%s\\%s.yuv"
#define ROOT_SEGFILE "%s\\%s.seg"
#define IOS_BINARY ios::binary
#define MKDIR "mkdir %s\\%2.2d"
#else
#define SUB_CMPFILE "%s/%2.2d/%s.cmp"
#define SUB_TRCFILE "%s/%2.2d/%s.trc"
#define SUB_YUVFILE	"%s/%2.2d/%s.yuv"
#define SUB_SEGFILE	"%s/%2.2d/%s.seg"
#define ENHN_SUB_CMPFILE "%s/%2.2d/%s_e.cmp"
#define ENHN_SUB_TRCFILE "%s/%2.2d/%s_e.trc"
#define ENHN_SUB_YUVFILE "%s/%2.2d/%s_e.yuv"
#define ENHN_SUB_SEGFILE "%s/%2.2d/%s_e.seg"
#define SUB_VDLFILE "%s/%2.2d/%s.vdl"	//for sprite i/o
#define SUB_PNTFILE "%s/%2.2d/%s.pnt"	//for sprite i/o
#define ROOT_YUVFILE "%s/%s.yuv"
#define ROOT_SEGFILE "%s/%s.seg"
//#define IOS_BINARY ios::bin
#define IOS_BINARY 0
#define MKDIR "mkdir %s/%2.2d"
#endif

#define BASE_LAYER 0
#define ENHN_LAYER 1

#define _FOR_GSSP_

Bool fileExistence (const Char* pchFile) 
{
	Bool ret = 1;
	FILE* pf = fopen (pchFile, "r");
	if (pf == NULL)
		ret = 0;
	else
		fclose (pf);
	return ret;
}

Void CSessionEncoder::createReconDir (UInt idx) const
{
	Char pchTmp [100];
	sprintf (pchTmp, SUB_YUVFILE, m_pchReconYUVDir, idx, m_pchPrefix);
	FILE* pf = fopen (pchTmp, "wb");
	if (pf == NULL) {
		sprintf (pchTmp, MKDIR, m_pchReconYUVDir, idx);
		system (pchTmp);																	
	}
	else
		fclose (pf);
}

Void CSessionEncoder::createCmpDir (UInt idx) const
{
	Char pchTmp [100];
	sprintf (pchTmp, SUB_CMPFILE, m_pchOutStrFiles, idx, m_pchPrefix);
	FILE* pf = fopen (pchTmp, "wb");
	if (pf == NULL) {
		sprintf (pchTmp, MKDIR, m_pchOutStrFiles, idx);
		system (pchTmp);																	
	}
	else
		fclose (pf);
}

CSessionEncoder::~CSessionEncoder ()
{
	delete [] m_rgvolmd [BASE_LAYER];
	delete [] m_rgvolmd [ENHN_LAYER];
	delete [] m_rgvopmd [BASE_LAYER];
	delete [] m_rgvopmd [ENHN_LAYER];

	// sprite 
	for (Int iVO = 0; iVO < m_iNumVO; iVO++) {
		delete [] m_ppstSrc [iVO];
		for (Int ifr = 0; ifr < m_iNumFrame; ifr++)
			delete [] m_pppstDst [iVO] [ifr];
		delete [] m_pppstDst [iVO];
	}
	delete [] m_ppstSrc;
	delete [] m_pppstDst;
}


CSessionEncoder::CSessionEncoder (
		// general info
		UInt uiFrmWidth, // frame width
		UInt uiFrmHeight, // frame height
		Int iFirstFrm, // first frame number
		Int iLastFrm, // last frame number
		Bool bNot8Bit, // NBIT: not 8-bit flag
		UInt uiQuantPrecision, // NBIT: quant precision
		UInt nBits, // NBIT: number of bits per pixel
		Int iFirstVO, // first VOP index
		Int iLastVO, // last VOP index
		const Bool* rgbSpatialScalability, // spatial scalability indicator
		const Int* rgiTemporalScalabilityType, // temporal scalability formation case // added by Sharp (98/02/09)
		const Int* rgiEnhancementType,	// enhancement_type for scalability // added by Sharp (98/02/09)
		UInt* rguiRateControl [], // rate control type
		UInt* rguiBudget [], // for rate control
		// for shape coding
		const AlphaUsage* rgfAlphaUsage, // alpha usage for each VOP.  0: binary, 1: 8-bit
		const Bool* rgbShapeOnly,		 // disable texture motion coding
		const Int* rgiBinaryAlphaTH,	 // threhold is shape size conversion
		const Int* rgiBinaryAlphaRR,	 // refrash rate for binary shape coding:	Added for error resilient mode by Toshiba(1997-11-14)
		const Bool* rgbNoCrChange,		 // no change of cr from mb to mb

		// motion estimation part for each VOP
		UInt** rguiSearchRange,  // motion search range
		Bool** rgbOriginalForME, // flag indicating whether use the original previous VOP for ME
		Bool** rgbAdvPredDisable,//no advanced MC (currenly = obmc, later = obmc + 8x8)
		// START: Complexity Estimation syntax support - Marc Mongenet (EPFL) - 17 Jun 1998
		Bool ** rgbComplexityEstimationDisable,
		Bool ** rgbOpaque,
		Bool ** rgbTransparent,
		Bool ** rgbIntraCAE,
		Bool ** rgbInterCAE,
		Bool ** rgbNoUpdate,
		Bool ** rgbUpsampling,
		Bool ** rgbIntraBlocks,
		Bool ** rgbInterBlocks,
		Bool ** rgbInter4vBlocks,
		Bool ** rgbNotCodedBlocks,
		Bool ** rgbDCTCoefs,
		Bool ** rgbDCTLines,
		Bool ** rgbVLCSymbols,
		Bool ** rgbVLCBits,
		Bool ** rgbAPM,
		Bool ** rgbNPM,
		Bool ** rgbInterpolateMCQ,
		Bool ** rgbForwBackMCQ,
		Bool ** rgbHalfpel2,
		Bool ** rgbHalfpel4,
		// END: Complexity Estimation syntax support

		// START: VOL Control Parameters
		UInt ** rguiVolControlParameters,
		UInt ** rguiChromaFormat,
		UInt ** rguiLowDelay,
		UInt ** rguiVBVParams,
		UInt ** rguiBitRate,
		UInt ** rguiVbvBufferSize,
		UInt ** rguiVbvBufferOccupany,
		// END: VOL Control Parameters

		Double** rgdFrameFrequency,

		// interlace coding
		Bool** rgbInterlacedCoding, // interlace flag
		Bool** rgbTopFieldFirst,    // top field first flag
        Bool** rgbAlternateScan,    // alternate scan flag

		Int** rgiDirectModeRadius,	// Direct mode search radius

		// motion vector file I/O
		Int** rgiMVFileUsage,
		Char*** pchMVFileName,
		// major syntax mode
		Int**	rgbVPBitTh,			// Bit threshold for video packet spacing control
		Bool** rgbDataPartitioning,  //data partitioning
		Bool** rgbReversibleVlc, //reversible VLC

		// for texture coding
		Quantizer** rgfQuant, // quantizer selection, either H.263 or MPEG
		Bool** rgbLoadIntraMatrix, // load user-defined intra Q-Matrix
		Int*** rgppiIntraQuantizerMatrix, // Intra Q-Matrix
		Bool** rgbLoadInterMatrix, // load user-defined inter Q-Matrix
		Int*** rgppiInterQuantizerMatrix, // Inter Q-Matrix
		Int** rgiIntraDCSwitchingThr,		//threshold to code dc as ac when pred. is on
		Int** rgiStepI, // I-VOP quantization stepsize
		Int** rgiStepP, // P-VOP quantization stepsize
		Bool**	rgbLoadIntraMatrixAlpha,
		Int*** rgppiIntraQuantizerMatrixAlpha,
		Bool**	rgbLoadInterMatrixAlpha,
		Int*** rgppiInterQuantizerMatrixAlpha,
		Int** rgiStepIAlpha, // I-VOP quantization stepsize for Alpha
		Int** rgiStepPAlpha, // P-VOP quantization stepsize for Alpha
		Int** rgiStepBAlpha, // P-VOP quantization stepsize for Alpha
		Int** rgbNoAlphaQuantUpdate, // discouple gray quant update with tex. quant
		Int** rgiStepB, // quantization stepsize for B-VOP
		const Int* rgiNumOfBbetweenPVOP,		// no of B-VOPs between P-VOPs
		const Int* rgiNumOfPbetweenIVOP,		// no of P-VOPs between I-VOPs
//added to encode GOVheader by SONY 980212
		const Int* rgiGOVperiod,
//980212
		const Bool* rgbDeblockFilterDisable, //deblocing filter disable
		const Bool *rgbAllowSkippedPMBs,

		// file information
		const Char* pchPrefix, // prefix name of the movie
		const Char* pchBmpFiles, // bmp file directory location
		const ChromType* rgfChrType, // input chrominance type. 0 - 4:4:4, 1 - 4:2:2, 0 - 4:2:0
		const Char* pchOutBmpFiles, // quantized frame file directory
		const Char* pchOutStrFiles, // output bitstream file
		const Int* rgiTemporalRate, // temporal subsampling rate
		const Int* rgiEnhnTemporalRate, // temporal subsampling rate // added by Sharp (98/02/09)
		
		// statistics dumping options
		const Bool* rgbDumpMB,
		const Bool* rgbTrace,

		// rounding control
		const Bool* rgbRoundingControlDisable,
		const Int* rgiInitialRoundingType,
		// for sprite info
		const UInt* rguiSpriteUsage, // sprite usage
		const UInt* rguiWarpingAccuracy, // warping accuracy
		const Int* rgNumOfPnts, // number of points for sprite, 0 for stationary and -1 for no sprite
		const Char* pchSptDir, // sprite directory
		const Char* pchSptPntDir, // sprite point file
		SptMode *pSpriteMode,		// sprite reconstruction mode, i.e. Low-latency-sprite-enable 
		Int iSpatialOption,