/* * jrevdct.c * * Copyright (C) 1991, 1992, 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 the basic inverse-DCT transformation subroutine. * * 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 "dct.h" /* We assume that right shift corresponds to signed division by 2 with * rounding towards minus infinity. This is correct for typical "arithmetic * shift" instructions that shift in copies of the sign bit. But some * C compilers implement >> with an unsigned shift. For these machines you * must define RIGHT_SHIFT_IS_UNSIGNED. * RIGHT_SHIFT provides a proper signed right shift of an INT32 quantity. * It is only applied with constant shift counts. SHIFT_TEMPS must be * included in the variables of any routine using RIGHT_SHIFT. */ #ifdef RIGHT_SHIFT_IS_UNSIGNED #define SHIFT_TEMPS INT32 shift_temp; #define RIGHT_SHIFT(x,shft) \ ((shift_temp = (x)) < 0 ? \ (shift_temp >> (shft)) | ((~((INT32) 0)) << (32-(shft))) : \ (shift_temp >> (shft))) #else #define SHIFT_TEMPS #define RIGHT_SHIFT(x,shft) ((x) >> (shft)) #endif /* * This routine is specialized to the case DCTSIZE = 8. */ #if DCTSIZE != 8 Sorry, this code only copes with 8x8 DCTs. /* deliberate syntax err */ #endif /* * A 2-D IDCT can be done by 1-D IDCT on each row followed by 1-D IDCT * 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. * * The poop on this scaling stuff is as follows: * * Each 1-D IDCT step produces outputs which are a factor of sqrt(N) * larger than the true IDCT 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 IDCT, * because the y0 and y4 inputs need not be divided by sqrt(N). * * 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 long 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. (To scale up 12-bit sample data further, an * intermediate INT32 array would be needed.) * * 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. */ #ifdef EIGHT_BIT_SAMPLES #define CONST_BITS 13 #define PASS1_BITS 2 #else #define CONST_BITS 13 #define PASS1_BITS 1 /* lose a little precision to avoid overflow */ #endif #define ONE ((INT32) 1) #define CONST_SCALE (ONE << CONST_BITS) /* Convert a positive real constant to an integer scaled by CONST_SCALE. */ #define FIX(x) ((INT32) ((x) * CONST_SCALE + 0.5)) /* 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 ((INT32) 2446) /* FIX(0.298631336) */ #define FIX_0_390180644 ((INT32) 3196) /* FIX(0.390180644) */ #define FIX_0_541196100 ((INT32) 4433) /* FIX(0.541196100) */ #define FIX_0_765366865 ((INT32) 6270) /* FIX(0.765366865) */ #define FIX_0_899976223 ((INT32) 7373) /* FIX(0.899976223) */ #define FIX_1_175875602 ((INT32) 9633) /* FIX(1.175875602) */ #define FIX_1_501321110 ((INT32) 12299) /* FIX(1.501321110) */ #define FIX_1_847759065 ((INT32) 15137) /* FIX(1.847759065) */ #define FIX_1_961570560 ((INT32) 16069) /* FIX(1.961570560) */ #define FIX_2_053119869 ((INT32) 16819) /* FIX(2.053119869) */ #define FIX_2_562915447 ((INT32) 20995) /* FIX(2.562915447) */ #define FIX_3_072711026 ((INT32) 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 /* 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) RIGHT_SHIFT((x) + (ONE << ((n)-1)), n) /* Multiply an INT32 variable by an INT32 constant to yield an INT32 result. * For 8-bit samples with the recommended scaling, all the variable * and constant values involved are no more than 16 bits wide, so a * 16x16->32 bit multiply can be used instead of a full 32x32 multiply; * this provides a useful speedup on many machines. * There is no way to specify a 16x16->32 multiply in portable C, but * some C compilers will do the right thing if you provide the correct * combination of casts. * NB: for 12-bit samples, a full 32-bit multiplication will be needed. */ #ifdef EIGHT_BIT_SAMPLES #ifdef SHORTxSHORT_32 /* may work if 'int' is 32 bits */ #define MULTIPLY(var,const) (((INT16) (var)) * ((INT16) (const))) #endif #ifdef SHORTxLCONST_32 /* known to work with Microsoft C 6.0 */ #define MULTIPLY(var,const) (((INT16) (var)) * ((INT32) (const))) #endif #endif #ifndef MULTIPLY /* default definition */ #define MULTIPLY(var,const) ((var) * (const)) #endif /* * Perform the inverse DCT on one block of coefficients. */ void gst_idct_int_idct (DCTBLOCK data) { INT32 tmp0, tmp1, tmp2, tmp3; INT32 tmp10, tmp11, tmp12, tmp13; INT32 z1, z2, z3, z4, z5; register DCTELEM *dataptr; int rowctr; SHIFT_TEMPS /* Pass 1: process rows. */ /* Note results are scaled up by sqrt(8) compared to a true IDCT; */ /* furthermore, we scale the results by 2**PASS1_BITS. */ dataptr = data; for (rowctr = DCTSIZE-1; rowctr >= 0; rowctr--) { /* Due to quantization, we will usually find that many of the input * coefficients are zero, especially the AC terms. We can exploit this * by short-circuiting the IDCT calculation for any row in which all * the AC terms are zero. In that case each output is equal to the * DC coefficient (with scale factor as needed). * With typical images and quantization tables, half or more of the * row DCT calculations can be simplified this way. */ if ((dataptr[1] | dataptr[2] | dataptr[3] | dataptr[4] | dataptr[5] | dataptr[6] | dataptr[7]) == 0) { /* AC terms all zero */ DCTELEM dcval = (DCTELEM) (dataptr[0] << PASS1_BITS); dataptr[0] = dcval; dataptr[1] = dcval; dataptr[2] = dcval; dataptr[3] = dcval; dataptr[4] = dcval; dataptr[5] = dcval; dataptr[6] = dcval; dataptr[7] = dcval; dataptr += DCTSIZE; /* advance pointer to next row */ continue; } /* Even part: reverse the even part of the forward DCT. */ /* The rotator is sqrt(2)*c(-6). */ z2 = (INT32) dataptr[2]; z3 = (INT32) dataptr[6]; z1 = MULTIPLY(z2 + z3, FIX_0_541196100); tmp2 = z1 + MULTIPLY(z3, - FIX_1_847759065); tmp3 = z1 + MULTIPLY(z2, FIX_0_765366865); tmp0 = ((INT32) dataptr[0] + (INT32) dataptr[4]) << CONST_BITS; tmp1 = ((INT32) dataptr[0] - (INT32) dataptr[4]) << CONST_BITS; tmp10 = tmp0 + tmp3; tmp13 = tmp0 - tmp3; tmp11 = tmp1 + tmp2; tmp12 = tmp1 - tmp2; /* Odd part per figure 8; the matrix is unitary and hence its * transpose is its inverse. i0..i3 are y7,y5,y3,y1 respectively. */ tmp0 = (INT32) dataptr[7]; tmp1 = (INT32) dataptr[5]; tmp2 = (INT32) dataptr[3]; tmp3 = (INT32) dataptr[1]; z1 = tmp0 + tmp3; z2 = tmp1 + tmp2; z3 = tmp0 + tmp2; z4 = tmp1 + tmp3; z5 = MULTIPLY(z3 + z4, FIX_1_175875602); /* sqrt(2) * c3 */ tmp0 = MULTIPLY(tmp0, FIX_0_298631336); /* sqrt(2) * (-c1+c3+c5-c7) */ tmp1 = MULTIPLY(tmp1, FIX_2_053119869); /* sqrt(2) * ( c1+c3-c5+c7) */ tmp2 = MULTIPLY(tmp2, FIX_3_072711026); /* sqrt(2) * ( c1+c3+c5-c7) */ tmp3 = MULTIPLY(tmp3, FIX_1_501321110); /* sqrt(2) * ( c1+c3-c5-c7) */ z1 = MULTIPLY(z1, - FIX_0_899976223); /* sqrt(2) * (c7-c3) */ z2 = MULTIPLY(z2, - FIX_2_562915447); /* sqrt(2) * (-c1-c3) */ z3 = MULTIPLY(z3, - FIX_1_961570560); /* sqrt(2) * (-c3-c5) */ z4 = MULTIPLY(z4, - FIX_0_390180644); /* sqrt(2) * (c5-c3) */ z3 += z5; z4 += z5; tmp0 += z1 + z3; tmp1 += z2 + z4; tmp2 += z2 + z3; tmp3 += z1 + z4; /* Final output stage: inputs are tmp10..tmp13, tmp0..tmp3 */ dataptr[0] = (DCTELEM) DESCALE(tmp10 + tmp3, CONST_BITS-PASS1_BITS); dataptr[7] = (DCTELEM) DESCALE(tmp10 - tmp3, CONST_BITS-PASS1_BITS); dataptr[1] = (DCTELEM) DESCALE(tmp11 + tmp2, CONST_BITS-PASS1_BITS); dataptr[6] = (DCTELEM) DESCALE(tmp11 - tmp2, CONST_BITS-PASS1_BITS); dataptr[2] = (DCTELEM) DESCALE(tmp12 + tmp1, CONST_BITS-PASS1_BITS); dataptr[5] = (DCTELEM) DESCALE(tmp12 - tmp1, CONST_BITS-PASS1_BITS); dataptr[3] = (DCTELEM) DESCALE(tmp13 + tmp0, CONST_BITS-PASS1_BITS); dataptr[4] = (DCTELEM) DESCALE(tmp13 - tmp0, CONST_BITS-PASS1_BITS); dataptr += DCTSIZE; /* advance pointer to next row */ } /* Pass 2: process columns. */ /* Note that we must descale the results by a factor of 8 == 2**3, */ /* and also undo the PASS1_BITS scaling. */ dataptr = data; for (rowctr = DCTSIZE-1; rowctr >= 0; rowctr--) { /* Columns of zeroes can be exploited in the same way as we did with rows. * However, the row calculation has created many nonzero AC terms, so the * simplification applies less often (typically 5% to 10% of the time). * On machines with very fast multiplication, it's possible that the * test takes more time than it's worth. In that case this section * may be commented out. */ #ifndef NO_ZERO_COLUMN_TEST if ((dataptr[DCTSIZE*1] | dataptr[DCTSIZE*2] | dataptr[DCTSIZE*3] | dataptr[DCTSIZE*4] | dataptr[DCTSIZE*5] | dataptr[DCTSIZE*6] | dataptr[DCTSIZE*7]) == 0) { /* AC terms all zero */ DCTELEM dcval = (DCTELEM) DESCALE((INT32) dataptr[0], PASS1_BITS+3); dataptr[DCTSIZE*0] = dcval; dataptr[DCTSIZE*1] = dcval; dataptr[DCTSIZE*2] = dcval; dataptr[DCTSIZE*3] = dcval; dataptr[DCTSIZE*4] = dcval; dataptr[DCTSIZE*5] = dcval; dataptr[DCTSIZE*6] = dcval; dataptr[DCTSIZE*7] = dcval; dataptr++; /* advance pointer to next column */ continue; } #endif /* Even part: reverse the even part of the forward DCT. */ /* The rotator is sqrt(2)*c(-6). */ z2 = (INT32) dataptr[DCTSIZE*2]; z3 = (INT32) dataptr[DCTSIZE*6]; z1 = MULTIPLY(z2 + z3, FIX_0_541196100); tmp2 = z1 + MULTIPLY(z3, - FIX_1_847759065); tmp3 = z1 + MULTIPLY(z2, FIX_0_765366865); tmp0 = ((INT32) dataptr[DCTSIZE*0] + (INT32) dataptr[DCTSIZE*4]) << CONST_BITS; tmp1 = ((INT32) dataptr[DCTSIZE*0] - (INT32) dataptr[DCTSIZE*4]) << CONST_BITS; tmp10 = tmp0 + tmp3; tmp13 = tmp0 - tmp3; tmp11 = tmp1 + tmp2; tmp12 = tmp1 - tmp2; /* Odd part per figure 8; the matrix is unitary and hence its * transpose is its inverse. i0..i3 are y7,y5,y3,y1 respectively. */ tmp0 = (INT32) dataptr[DCTSIZE*7]; tmp1 = (INT32) dataptr[DCTSIZE*5]; tmp2 = (INT32) dataptr[DCTSIZE*3]; tmp3 = (INT32) dataptr[DCTSIZE*1]; z1 = tmp0 + tmp3; z2 = tmp1 + tmp2; z3 = tmp0 + tmp2; z4 = tmp1 + tmp3; z5 = MULTIPLY(z3 + z4, FIX_1_175875602); /* sqrt(2) * c3 */ tmp0 = MULTIPLY(tmp0, FIX_0_298631336); /* sqrt(2) * (-c1+c3+c5-c7) */ tmp1 = MULTIPLY(tmp1, FIX_2_053119869); /* sqrt(2) * ( c1+c3-c5+c7) */ tmp2 = MULTIPLY(tmp2, FIX_3_072711026); /* sqrt(2) * ( c1+c3+c5-c7) */ tmp3 = MULTIPLY(tmp3, FIX_1_501321110); /* sqrt(2) * ( c1+c3-c5-c7) */ z1 = MULTIPLY(z1, - FIX_0_899976223); /* sqrt(2) * (c7-c3) */ z2 = MULTIPLY(z2, - FIX_2_562915447); /* sqrt(2) * (-c1-c3) */ z3 = MULTIPLY(z3, - FIX_1_961570560); /* sqrt(2) * (-c3-c5) */ z4 = MULTIPLY(z4, - FIX_0_390180644); /* sqrt(2) * (c5-c3) */ z3 += z5; z4 += z5; tmp0 += z1 + z3; tmp1 += z2 + z4; tmp2 += z2 + z3; tmp3 += z1 + z4; /* Final output stage: inputs are tmp10..tmp13, tmp0..tmp3 */ dataptr[DCTSIZE*0] = (DCTELEM) DESCALE(tmp10 + tmp3, CONST_BITS+PASS1_BITS+3); dataptr[DCTSIZE*7] = (DCTELEM) DESCALE(tmp10 - tmp3, CONST_BITS+PASS1_BITS+3); dataptr[DCTSIZE*1] = (DCTELEM) DESCALE(tmp11 + tmp2, CONST_BITS+PASS1_BITS+3); dataptr[DCTSIZE*6] = (DCTELEM) DESCALE(tmp11 - tmp2, CONST_BITS+PASS1_BITS+3); dataptr[DCTSIZE*2] = (DCTELEM) DESCALE(tmp12 + tmp1, CONST_BITS+PASS1_BITS+3); dataptr[DCTSIZE*5] = (DCTELEM) DESCALE(tmp12 - tmp1, CONST_BITS+PASS1_BITS+3); dataptr[DCTSIZE*3] = (DCTELEM) DESCALE(tmp13 + tmp0, CONST_BITS+PASS1_BITS+3); dataptr[DCTSIZE*4] = (DCTELEM) DESCALE(tmp13 - tmp0, CONST_BITS+PASS1_BITS+3); dataptr++; /* advance pointer to next column */ } }