jfdctint.c

jpeg压缩C代码,包括库的源代码和一个测试程序的源代码

/*
* jfdctint.c
*
* Copyright (C) 1991-1996, Thomas G. Lane.
* This file is part of the Independent JPEG Group's software.
* For conditions of distribution and use, see the accompanying README file.
*
* This file contains a slow-but-accurate integer implementation of the
* forward DCT (Discrete Cosine Transform).
*
* A 2-D DCT can be done by 1-D DCT on each row followed by 1-D DCT
* on each column.  Direct algorithms are also available, but they are
* much more complex and seem not to be any faster when reduced to code.
*
* This implementation is based on an algorithm described in
*   C. Loeffler, A. Ligtenberg and G. Moschytz, "Practical Fast 1-D DCT
*   Algorithms with 11 Multiplications", Proc. Int'l. Conf. on Acoustics,
*   Speech, and Signal Processing 1989 (ICASSP '89), pp. 988-991.
* The primary algorithm described there uses 11 multiplies and 29 adds.
* We use their alternate method with 12 multiplies and 32 adds.
* The advantage of this method is that no data path contains more than one
* multiplication; this allows a very simple and accurate implementation in
* scaled fixed-point arithmetic, with a minimal number of shifts.
*/

#include "jpeg.h"

/*
* The poop on this scaling stuff is as follows:
*
* Each 1-D DCT step produces outputs which are a factor of sqrt(N)
* larger than the true DCT outputs.  The final outputs are therefore
* a factor of N larger than desired; since N=8 this can be cured by
* a simple right shift at the end of the algorithm.  The advantage of
* this arrangement is that we save two multiplications per 1-D DCT,
* because the y0 and y4 outputs need not be divided by sqrt(N).
* In the IJG code, this factor of 8 is removed by the quantization step
* (in jcdctmgr.c), NOT in this module.
*
* We have to do addition and subtraction of the integer inputs, which
* is no problem, and multiplication by fractional constants, which is
* a problem to do in integer arithmetic.  We multiply all the constants
* by CONST_SCALE and convert them to integer constants (thus retaining
* CONST_BITS bits of precision in the constants).  After doing a
* multiplication we have to divide the product by CONST_SCALE, with proper
* rounding, to produce the correct output.  This division can be done
* cheaply as a right shift of CONST_BITS bits.  We postpone shifting
* as int as possible so that partial sums can be added together with
* full fractional precision.
*
* The outputs of the first pass are scaled up by PASS1_BITS bits so that
* they are represented to better-than-integral precision.  These outputs
* require BITS_IN_JSAMPLE + PASS1_BITS + 3 bits; this fits in a 16-bit word
* with the recommended scaling.  (For 12-bit sample data, the intermediate
* array is INT32 anyway.)
*
* To avoid overflow of the 32-bit intermediate results in pass 2, we must
* have BITS_IN_JSAMPLE + CONST_BITS + PASS1_BITS <= 26.  Error analysis
* shows that the values given below are the most effective.
*/

#define CONST_BITS  13
#define PASS1_BITS  2


/* Some C compilers fail to reduce "FIX(constant)" at compile time, thus
* causing a lot of useless floating-point operations at run time.
* To get around this we use the following pre-calculated constants.
* If you change CONST_BITS you may want to add appropriate values.
* (With a reasonable C compiler, you can just rely on the FIX() macro...)
*/

#if CONST_BITS == 13
#define FIX_0_298631336  ((int)  2446) /* FIX(0.298631336) */
#define FIX_0_390180644  ((int)  3196) /* FIX(0.390180644) */
#define FIX_0_541196100  ((int)  4433) /* FIX(0.541196100) */
#define FIX_0_765366865  ((int)  6270) /* FIX(0.765366865) */
#define FIX_0_899976223  ((int)  7373) /* FIX(0.899976223) */
#define FIX_1_175875602  ((int)  9633) /* FIX(1.175875602) */
#define FIX_1_501321110  ((int)  12299) /* FIX(1.501321110) */
#define FIX_1_847759065  ((int)  15137) /* FIX(1.847759065) */
#define FIX_1_961570560  ((int)  16069) /* FIX(1.961570560) */
#define FIX_2_053119869  ((int)  16819) /* FIX(2.053119869) */
#define FIX_2_562915447  ((int)  20995) /* FIX(2.562915447) */
#define FIX_3_072711026  ((int)  25172) /* FIX(3.072711026) */
#else
#define FIX_0_298631336  FIX(0.298631336)
#define FIX_0_390180644  FIX(0.390180644)
#define FIX_0_541196100  FIX(0.541196100)
#define FIX_0_765366865  FIX(0.765366865)
#define FIX_0_899976223  FIX(0.899976223)
#define FIX_1_175875602  FIX(1.175875602)
#define FIX_1_501321110  FIX(1.501321110)
#define FIX_1_847759065  FIX(1.847759065)
#define FIX_1_961570560  FIX(1.961570560)
#define FIX_2_053119869  FIX(2.053119869)
#define FIX_2_562915447  FIX(2.562915447)
#define FIX_3_072711026  FIX(3.072711026)
#endif


/* Convert a positive real constant to an integer scaled by CONST_SCALE.
* Caution: some C compilers fail to reduce "FIX(constant)" at compile time,
* thus causing a lot of useless floating-point operations at run time.
*/

#define FIX(x)	((int) ((x) * (1 << CONST_BITS) + 0.5))

#define RIGHT_SHIFT(x,shft)	((x) >> (shft))
/* Descale and correctly round an INT32 value that's scaled by N bits.
* We assume RIGHT_SHIFT rounds towards minus infinity, so adding
* the fudge factor is correct for either sign of X.
*/

#define DESCALE(x,n)  ((short)RIGHT_SHIFT((x) + (1 << ((n)-1)), n))

/*
* Perform an integer forward DCT on one block of samples.
*/
void jpeg_fdct_islow(short * block)
{
	int tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
	int tmp10, tmp11, tmp12, tmp13;
	int z1, z2, z3, z4, z5;
	short *blkptr;
	short *dataptr;
	short data[64];
	int i;
	
	/* Pass 1: process rows. */
	/* Note results are scaled up by sqrt(8) compared to a true DCT; */
	/* furthermore, we scale the results by 2**PASS1_BITS. */
	
	dataptr = data;
	blkptr = block;
	for (i = 0; i < 8; i++) 
	{
		tmp0 = blkptr[0] + blkptr[7];
		tmp7 = blkptr[0] - blkptr[7];
		tmp1 = blkptr[1] + blkptr[6];
		tmp6 = blkptr[1] - blkptr[6];
		tmp2 = blkptr[2] + blkptr[5];
		tmp5 = blkptr[2] - blkptr[5];
		tmp3 = blkptr[3] + blkptr[4];
		tmp4 = blkptr[3] - blkptr[4];
		
		/* Even part per LL&M figure 1 --- note that published figure is faulty;
		* rotator "sqrt(2)*c1" should be "sqrt(2)*c6".
		*/
		
		tmp10 = tmp0 + tmp3;
		tmp13 = tmp0 - tmp3;
		tmp11 = tmp1 + tmp2;
		tmp12 = tmp1 - tmp2;
		
		dataptr[0] = (tmp10 + tmp11) << PASS1_BITS;
		dataptr[4] = (tmp10 - tmp11) << PASS1_BITS;
		
		z1 = (tmp12 + tmp13) * FIX_0_541196100;
		dataptr[2] = DESCALE(z1 + tmp13 * FIX_0_765366865, CONST_BITS - PASS1_BITS);
		dataptr[6] = DESCALE(z1 + tmp12 * (-FIX_1_847759065), CONST_BITS - PASS1_BITS);
		
		/* Odd part per figure 8 --- note paper omits factor of sqrt(2).
		* cK represents cos(K*pi/16).
		* i0..i3 in the paper are tmp4..tmp7 here.
		*/
		
		z1 = tmp4 + tmp7;
		z2 = tmp5 + tmp6;
		z3 = tmp4 + tmp6;
		z4 = tmp5 + tmp7;
		z5 = (z3 + z4) * FIX_1_175875602; /* sqrt(2) * c3 */
		
		tmp4 *= FIX_0_298631336; /* sqrt(2) * (-c1+c3+c5-c7) */
		tmp5 *= FIX_2_053119869; /* sqrt(2) * ( c1+c3-c5+c7) */
		tmp6 *= FIX_3_072711026; /* sqrt(2) * ( c1+c3+c5-c7) */
		tmp7 *= FIX_1_501321110; /* sqrt(2) * ( c1+c3-c5-c7) */
		z1 *= -FIX_0_899976223; /* sqrt(2) * (c7-c3) */
		z2 *= -FIX_2_562915447; /* sqrt(2) * (-c1-c3) */
		z3 *= -FIX_1_961570560; /* sqrt(2) * (-c3-c5) */
		z4 *= -FIX_0_390180644; /* sqrt(2) * (c5-c3) */
		
		z3 += z5;
		z4 += z5;
		
		dataptr[7] = DESCALE(tmp4 + z1 + z3, CONST_BITS - PASS1_BITS);
		dataptr[5] = DESCALE(tmp5 + z2 + z4, CONST_BITS - PASS1_BITS);
		dataptr[3] = DESCALE(tmp6 + z2 + z3, CONST_BITS - PASS1_BITS);
		dataptr[1] = DESCALE(tmp7 + z1 + z4, CONST_BITS - PASS1_BITS);
		
		dataptr += 8; /* advance pointer to next row */
		blkptr += 8;
	}
	
	/* Pass 2: process columns.
	* We remove the PASS1_BITS scaling, but leave the results scaled up
	* by an overall factor of 8.
	*/
	
	dataptr = data;
	for (i = 0; i < 8; i++) 
	{
		tmp0 = dataptr[0] + dataptr[56];
		tmp7 = dataptr[0] - dataptr[56];
		tmp1 = dataptr[8] + dataptr[48];
		tmp6 = dataptr[8] - dataptr[48];
		tmp2 = dataptr[16] + dataptr[40];
		tmp5 = dataptr[16] - dataptr[40];
		tmp3 = dataptr[24] + dataptr[32];
		tmp4 = dataptr[24] - dataptr[32];
		
		/* Even part per LL&M figure 1 --- note that published figure is faulty;
		* rotator "sqrt(2)*c1" should be "sqrt(2)*c6".
		*/
		
		tmp10 = tmp0 + tmp3;
		tmp13 = tmp0 - tmp3;
		tmp11 = tmp1 + tmp2;
		tmp12 = tmp1 - tmp2;
		
		dataptr[0] = DESCALE(tmp10 + tmp11, PASS1_BITS);
		dataptr[32] = DESCALE(tmp10 - tmp11, PASS1_BITS);
		
		z1 = (tmp12 + tmp13) * FIX_0_541196100;
		dataptr[16] = DESCALE(z1 + tmp13 * FIX_0_765366865, CONST_BITS + PASS1_BITS);
		dataptr[48] = DESCALE(z1 + tmp12 * (-FIX_1_847759065), CONST_BITS + PASS1_BITS);
		
		/* Odd part per figure 8 --- note paper omits factor of sqrt(2).
		* cK represents cos(K*pi/16).
		* i0..i3 in the paper are tmp4..tmp7 here.
		*/
		
		z1 = tmp4 + tmp7;
		z2 = tmp5 + tmp6;
		z3 = tmp4 + tmp6;
		z4 = tmp5 + tmp7;
		z5 = (z3 + z4) * FIX_1_175875602; /* sqrt(2) * c3 */
		
		tmp4 *= FIX_0_298631336; /* sqrt(2) * (-c1+c3+c5-c7) */
		tmp5 *= FIX_2_053119869; /* sqrt(2) * ( c1+c3-c5+c7) */
		tmp6 *= FIX_3_072711026; /* sqrt(2) * ( c1+c3+c5-c7) */
		tmp7 *= FIX_1_501321110; /* sqrt(2) * ( c1+c3-c5-c7) */
		z1 *= -FIX_0_899976223; /* sqrt(2) * (c7-c3) */
		z2 *= -FIX_2_562915447; /* sqrt(2) * (-c1-c3) */