jj2000.j2k.quantization.dequantizer.StdDequantizer Maven / Gradle / Ivy
Show all versions of jai-imageio-jpeg2000 Show documentation
/*
* $RCSfile: StdDequantizer.java,v $
* $Revision: 1.1 $
* $Date: 2005/02/11 05:02:19 $
* $State: Exp $
*
* Class: StdDequantizer
*
* Description: Scalar deadzone dequantizer that returns integers
* or floats.
* This is a merger of the ScalarDZDeqInt and
* ScalarDZDeqFloat classes by Joel Askelof and Diego
* Santa Cruz.
*
*
*
* COPYRIGHT:
*
* This software module was originally developed by Raphaël Grosbois and
* Diego Santa Cruz (Swiss Federal Institute of Technology-EPFL); Joel
* Askelöf (Ericsson Radio Systems AB); and Bertrand Berthelot, David
* Bouchard, Félix Henry, Gerard Mozelle and Patrice Onno (Canon Research
* Centre France S.A) in the course of development of the JPEG2000
* standard as specified by ISO/IEC 15444 (JPEG 2000 Standard). This
* software module is an implementation of a part of the JPEG 2000
* Standard. Swiss Federal Institute of Technology-EPFL, Ericsson Radio
* Systems AB and Canon Research Centre France S.A (collectively JJ2000
* Partners) agree not to assert against ISO/IEC and users of the JPEG
* 2000 Standard (Users) any of their rights under the copyright, not
* including other intellectual property rights, for this software module
* with respect to the usage by ISO/IEC and Users of this software module
* or modifications thereof for use in hardware or software products
* claiming conformance to the JPEG 2000 Standard. Those intending to use
* this software module in hardware or software products are advised that
* their use may infringe existing patents. The original developers of
* this software module, JJ2000 Partners and ISO/IEC assume no liability
* for use of this software module or modifications thereof. No license
* or right to this software module is granted for non JPEG 2000 Standard
* conforming products. JJ2000 Partners have full right to use this
* software module for his/her own purpose, assign or donate this
* software module to any third party and to inhibit third parties from
* using this software module for non JPEG 2000 Standard conforming
* products. This copyright notice must be included in all copies or
* derivative works of this software module.
*
* Copyright (c) 1999/2000 JJ2000 Partners.
* */
package jj2000.j2k.quantization.dequantizer;
import jj2000.j2k.decoder.DecoderSpecs;
import jj2000.j2k.image.DataBlk;
import jj2000.j2k.image.DataBlkFloat;
import jj2000.j2k.image.DataBlkInt;
import jj2000.j2k.quantization.GuardBitsSpec;
import jj2000.j2k.quantization.QuantStepSizeSpec;
import jj2000.j2k.quantization.QuantTypeSpec;
import jj2000.j2k.wavelet.synthesis.SubbandSyn;
/**
* This class implements a scalar dequantizer with deadzone. The output can be
* either integer ('int') or floating-point ('float') data. The dequantization
* step sizes and other parameters are taken from a StdDequantizerParams
* class, which inherits from DequantizerParams.
*
* Sign magnitude representation is used (instead of two's complement) for
* the input data. The most significant bit is used for the sign (0 if
* positive, 1 if negative). Then the magnitude of the quantized coefficient
* is stored in the next most significat bits. The most significant magnitude
* bit corresponds to the most significant bit-plane and so on.
*
*
When reversible quantization is used, this class only converts between
* the sign-magnitude representation and the integer (or eventually
* fixed-point) output, since there is no true quantization.
*
*
The output data is fixed-point two's complement for 'int' output and
* floating-point for 'float' output. The type of output and the number number
* of fractional bits for 'int' output are defined at the constructor. Each
* component may have a different number of fractional bits.
*
*
The reconstruction levels used by the dequantizer are exactly what is
* received from the entropy decoder. It is assumed that the entropy decoder
* always returns codewords that are midways in the decoded intervals. In this
* way the dequantized values will always lie midways in the quantization
* intervals.
* */
public class StdDequantizer extends Dequantizer {
/** The quantizer type spec */
private QuantTypeSpec qts;
/** The quantizer step sizes spec */
private QuantStepSizeSpec qsss;
/** The number of guard bits spec */
private GuardBitsSpec gbs;
/** The decoding parameters of the dequantizer */
private StdDequantizerParams params;
/** The 'DataBlkInt' object used to request data, used when output data is
* not int */
private DataBlkInt inblk;
/** Type of the current output data */
private int outdtype;
/**
* Initializes the source of compressed data. And sets the number of range
* bits and fraction bits and receives the parameters for the dequantizer.
*
* @param src From where to obtain the quantized data.
*
* @param rb The number of "range bits" (bitdepth) for each component
* (must be the "range bits" of the un-transformed components). For a
* definition of "range bits" see the getNomRangeBits() method.
*
* @param qts The quantizer type spec
*
* @param qsss The dequantizer step sizes spec
*
* @see Dequantizer#getNomRangeBits
*
* @exception IllegalArgumentException Thrown if 'outdt' is neither
* TYPE_FLOAT nor TYPE_INT, or if 'param' specify reversible quantization
* and 'outdt' is not TYPE_INT or 'fp' has non-zero values, or if 'outdt'
* is TYPE_FLOAT and 'fp' has non-zero values.
* */
public StdDequantizer(CBlkQuantDataSrcDec src,int[] utrb,
DecoderSpecs decSpec){
super(src,utrb,decSpec);
if(utrb.length != src.getNumComps()){
throw new IllegalArgumentException("Invalid rb argument");
}
this.qsss = decSpec.qsss;
this.qts = decSpec.qts;
this.gbs = decSpec.gbs;
}
/**
* Returns the position of the fixed point in the output data for the
* specified component. This is the position of the least significant
* integral (i.e. non-fractional) bit, which is equivalent to the number
* of fractional bits. For instance, for fixed-point values with 2
* fractional bits, 2 is returned. For floating-point data this value does
* not apply and 0 should be returned. Position 0 is the position of the
* least significant bit in the data. If the output data is 'float' then 0
* is always returned.
*
*
Note: Fractional bits are no more supported by JJ2000.
*
* @param c The index of the component.
*
* @return The position of the fixed-point, which is the same as
* the number of fractional bits. For floating-point data 0 is
* returned.
* */
public int getFixedPoint(int c){
return 0;
}
/**
* Returns the specified code-block in the current tile for the specified
* component, as a copy (see below).
*
*
The returned code-block may be progressive, which is indicated by
* the 'progressive' variable of the returned 'DataBlk' object. If a
* code-block is progressive it means that in a later request to this
* method for the same code-block it is possible to retrieve data which is
* a better approximation, since meanwhile more data to decode for the
* code-block could have been received. If the code-block is not
* progressive then later calls to this method for the same code-block
* will return the exact same data values.
*
*
The data returned by this method is always a copy of the internal
* data of this object, if any, and it can be modified "in place" without
* any problems after being returned. The 'offset' of the returned data is
* 0, and the 'scanw' is the same as the code-block width. See the
* 'DataBlk' class.
*
* @param c The component for which to return the next code-block.
*
* @param m The vertical index of the code-block to return, in the
* specified subband.
*
* @param n The horizontal index of the code-block to return, in the
* specified subband.
*
* @param sb The subband in which the code-block to return is.
*
* @param cblk If non-null this object will be used to return the new
* code-block. If null a new one will be allocated and returned. If the
* "data" array of the object is non-null it will be reused, if possible,
* to return the data.
*
* @return The next code-block in the current tile for component 'n', or
* null if all code-blocks for the current tile have been returned.
*
* @see DataBlk
* */
public final DataBlk getCodeBlock(int c, int m, int n, SubbandSyn sb,
DataBlk cblk) {
return getInternCodeBlock(c,m,n,sb,cblk);
}
/**
* Returns the specified code-block in the current tile for the specified
* component (as a reference or copy).
*
*
The returned code-block may be progressive, which is indicated by
* the 'progressive' variable of the returned 'DataBlk'
* object. If a code-block is progressive it means that in a later request
* to this method for the same code-block it is possible to retrieve data
* which is a better approximation, since meanwhile more data to decode
* for the code-block could have been received. If the code-block is not
* progressive then later calls to this method for the same code-block
* will return the exact same data values.
*
*
The data returned by this method can be the data in the internal
* buffer of this object, if any, and thus can not be modified by the
* caller. The 'offset' and 'scanw' of the returned data can be
* arbitrary. See the 'DataBlk' class.
*
* @param c The component for which to return the next code-block.
*
* @param m The vertical index of the code-block to return, in the
* specified subband.
*
* @param n The horizontal index of the code-block to return, in the
* specified subband.
*
* @param sb The subband in which the code-block to return is.
*
* @param cblk If non-null this object will be used to return the new
* code-block. If null a new one will be allocated and returned. If the
* "data" array of the object is non-null it will be reused, if possible,
* to return the data.
*
* @return The next code-block in the current tile for component 'n', or
* null if all code-blocks for the current tile have been returned.
*
* @see DataBlk
* */
public final
DataBlk getInternCodeBlock(int c, int m, int n, SubbandSyn sb,
DataBlk cblk) {
// This method is declared final since getNextCodeBlock() relies on
// the actual implementation of this method.
int j,jmin,k;
int temp;
float step;
int shiftBits;
int magBits;
int[] outiarr,inarr;
float[] outfarr;
int w,h;
boolean reversible = qts.isReversible(tIdx,c);
boolean derived = qts.isDerived(tIdx,c);
StdDequantizerParams
params = (StdDequantizerParams)qsss.getTileCompVal(tIdx,c);
int G = ((Integer)gbs.getTileCompVal(tIdx,c)).intValue();
outdtype = cblk.getDataType();
if (reversible && outdtype!=DataBlk.TYPE_INT) {
throw new IllegalArgumentException("Reversible quantizations "+
"must use int data");
}
// To get compiler happy
outiarr = null;
outfarr = null;
inarr = null;
// Get source data and initialize output DataBlk object.
switch (outdtype) {
case DataBlk.TYPE_INT:
// With int data we can use the same DataBlk object to get the
// data from the source and return the dequantized data, and we
// can also work "in place" (i.e. same buffer).
cblk = src.getCodeBlock(c,m,n,sb,cblk);
// Input and output arrays are the same
outiarr = (int[]) cblk.getData();
break;
case DataBlk.TYPE_FLOAT:
// With float data we must use a different DataBlk objects to get
// the data from the source and to return the dequantized data.
inblk = (DataBlkInt) src.getInternCodeBlock(c,m,n,sb,inblk);
inarr = inblk.getDataInt();
if (cblk == null) {
cblk = new DataBlkFloat();
}
// Copy the attributes of the CodeBlock object
cblk.ulx = inblk.ulx;
cblk.uly = inblk.uly;
cblk.w = inblk.w;
cblk.h = inblk.h;
cblk.offset = 0;
cblk.scanw = cblk.w;
cblk.progressive = inblk.progressive;
// Get output data array and check its size
outfarr = (float[]) cblk.getData();
if (outfarr == null || outfarr.length < cblk.w*cblk.h) {
outfarr = new float[cblk.w*cblk.h];
cblk.setData(outfarr);
}
break;
}
magBits = sb.magbits;
// Calculate quantization step and number of magnitude bits
// depending on reversibility and derivedness and perform
// inverse quantization
if(reversible){
shiftBits=31-magBits;
// For int data Inverse quantization happens "in-place". The input
// array has an offset of 0 and scan width equal to the code-block
// width.
for (j=outiarr.length-1; j>=0; j--) {
temp = outiarr[j]; // input array is same as output one
outiarr[j]=(temp >= 0) ? (temp>>shiftBits) :
-((temp&0x7FFFFFFF)>>shiftBits);
}
}
else{// Not reversible
if(derived){
// Max resolution level
int mrl = src.getSynSubbandTree(getTileIdx(),c).resLvl;
step=params.nStep[0][0]*
(1L<<(rb[c]+sb.anGainExp+mrl-sb.level));
}
else{
step=params.nStep[sb.resLvl][sb.sbandIdx]*
(1L<<(rb[c]+sb.anGainExp));
}
shiftBits=31-magBits;
// Adjust step to the number of shiftBits
step /= (1<=0; j--) {
temp = outiarr[j]; // input array is same as output one
outiarr[j] = (int)(((float)((temp >= 0) ? temp :
-(temp&0x7FFFFFFF)))*step);
}
break;
case DataBlk.TYPE_FLOAT:
// For float data the inverse quantization can not happen
// "in-place".
w = cblk.w;
h = cblk.h;
for (j=w*h-1, k=inblk.offset+(h-1)*inblk.scanw+w-1,
jmin = w*(h-1); j>=0; jmin -= w) {
for (; j>=jmin; k--, j--) {
temp = inarr[k];
outfarr[j] = ((float)((temp >= 0) ? temp :
-(temp&0x7FFFFFFF)))*step;
}
// Jump to beggining of previous line in input
k -= inblk.scanw - w;
}
break;
}
}
// Return the output code-block
return cblk;
}
}