1557
|
1 /*
|
|
2 * Floating point AAN DCT
|
|
3 * Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at>
|
|
4 *
|
|
5 * This library is free software; you can redistribute it and/or
|
|
6 * modify it under the terms of the GNU Lesser General Public
|
|
7 * License as published by the Free Software Foundation; either
|
|
8 * version 2 of the License, or (at your option) any later version.
|
|
9 *
|
|
10 * This library is distributed in the hope that it will be useful,
|
|
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
13 * Lesser General Public License for more details.
|
|
14 *
|
|
15 * You should have received a copy of the GNU Lesser General Public
|
|
16 * License along with this library; if not, write to the Free Software
|
|
17 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
|
18 *
|
1558
|
19 * this implementation is based upon the IJG integer AAN DCT (see jfdctfst.c)
|
1557
|
20 */
|
|
21
|
|
22 /**
|
|
23 * @file faandct.c
|
|
24 * @brief
|
|
25 * Floating point AAN DCT
|
|
26 * @author Michael Niedermayer <michaelni@gmx.at>
|
|
27 */
|
|
28
|
|
29 #include <math.h>
|
|
30
|
|
31 #include "dsputil.h"
|
|
32 #include "faandct.h"
|
|
33
|
|
34 #define FLOAT float
|
|
35
|
|
36 //numbers generated by simple c code (not as accurate as they could be)
|
|
37 /*
|
|
38 for(i=0; i<8; i++){
|
|
39 printf("#define B%d %1.20llf\n", i, (long double)1.0/(cosl(i*acosl(-1.0)/(long double)16.0)*sqrtl(2)));
|
|
40 }
|
|
41 */
|
|
42 #define B0 1.00000000000000000000
|
|
43 #define B1 0.72095982200694791383 // (cos(pi*1/16)sqrt(2))^-1
|
|
44 #define B2 0.76536686473017954350 // (cos(pi*2/16)sqrt(2))^-1
|
|
45 #define B3 0.85043009476725644878 // (cos(pi*3/16)sqrt(2))^-1
|
|
46 #define B4 1.00000000000000000000 // (cos(pi*4/16)sqrt(2))^-1
|
|
47 #define B5 1.27275858057283393842 // (cos(pi*5/16)sqrt(2))^-1
|
|
48 #define B6 1.84775906502257351242 // (cos(pi*6/16)sqrt(2))^-1
|
|
49 #define B7 3.62450978541155137218 // (cos(pi*7/16)sqrt(2))^-1
|
|
50
|
|
51
|
|
52 #define A1 0.70710678118654752438 // cos(pi*4/16)
|
|
53 #define A2 0.54119610014619698435 // cos(pi*6/16)sqrt(2)
|
|
54 #define A5 0.38268343236508977170 // cos(pi*6/16)
|
|
55 #define A4 1.30656296487637652774 // cos(pi*2/16)sqrt(2)
|
|
56
|
|
57 static FLOAT postscale[64]={
|
|
58 B0*B0, B0*B1, B0*B2, B0*B3, B0*B4, B0*B5, B0*B6, B0*B7,
|
|
59 B1*B0, B1*B1, B1*B2, B1*B3, B1*B4, B1*B5, B1*B6, B1*B7,
|
|
60 B2*B0, B2*B1, B2*B2, B2*B3, B2*B4, B2*B5, B2*B6, B2*B7,
|
|
61 B3*B0, B3*B1, B3*B2, B3*B3, B3*B4, B3*B5, B3*B6, B3*B7,
|
|
62 B4*B0, B4*B1, B4*B2, B4*B3, B4*B4, B4*B5, B4*B6, B4*B7,
|
|
63 B5*B0, B5*B1, B5*B2, B5*B3, B5*B4, B5*B5, B5*B6, B5*B7,
|
|
64 B6*B0, B6*B1, B6*B2, B6*B3, B6*B4, B6*B5, B6*B6, B6*B7,
|
|
65 B7*B0, B7*B1, B7*B2, B7*B3, B7*B4, B7*B5, B7*B6, B7*B7,
|
|
66 };
|
|
67
|
|
68 void ff_faandct(DCTELEM * data)
|
|
69 {
|
|
70 FLOAT tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
|
|
71 FLOAT tmp10, tmp11, tmp12, tmp13;
|
|
72 FLOAT z1, z2, z3, z4, z5, z11, z13;
|
|
73 FLOAT temp[64];
|
|
74 int i;
|
|
75
|
1561
|
76 emms_c();
|
1557
|
77
|
|
78 for (i=0; i<8*8; i+=8) {
|
|
79 tmp0= data[0 + i] + data[7 + i];
|
|
80 tmp7= data[0 + i] - data[7 + i];
|
|
81 tmp1= data[1 + i] + data[6 + i];
|
|
82 tmp6= data[1 + i] - data[6 + i];
|
|
83 tmp2= data[2 + i] + data[5 + i];
|
|
84 tmp5= data[2 + i] - data[5 + i];
|
|
85 tmp3= data[3 + i] + data[4 + i];
|
|
86 tmp4= data[3 + i] - data[4 + i];
|
|
87
|
|
88 tmp10= tmp0 + tmp3;
|
|
89 tmp13= tmp0 - tmp3;
|
|
90 tmp11= tmp1 + tmp2;
|
|
91 tmp12= tmp1 - tmp2;
|
|
92
|
|
93 temp[0 + i]= tmp10 + tmp11;
|
|
94 temp[4 + i]= tmp10 - tmp11;
|
|
95
|
|
96 z1= (tmp12 + tmp13)*A1;
|
|
97 temp[2 + i]= tmp13 + z1;
|
|
98 temp[6 + i]= tmp13 - z1;
|
|
99
|
|
100 tmp10= tmp4 + tmp5;
|
|
101 tmp11= tmp5 + tmp6;
|
|
102 tmp12= tmp6 + tmp7;
|
|
103
|
|
104 z5= (tmp10 - tmp12) * A5;
|
|
105 z2= tmp10*A2 + z5;
|
|
106 z4= tmp12*A4 + z5;
|
|
107 z3= tmp11*A1;
|
|
108
|
|
109 z11= tmp7 + z3;
|
|
110 z13= tmp7 - z3;
|
|
111
|
|
112 temp[5 + i]= z13 + z2;
|
|
113 temp[3 + i]= z13 - z2;
|
|
114 temp[1 + i]= z11 + z4;
|
|
115 temp[7 + i]= z11 - z4;
|
|
116 }
|
|
117
|
|
118 for (i=0; i<8; i++) {
|
|
119 tmp0= temp[8*0 + i] + temp[8*7 + i];
|
|
120 tmp7= temp[8*0 + i] - temp[8*7 + i];
|
|
121 tmp1= temp[8*1 + i] + temp[8*6 + i];
|
|
122 tmp6= temp[8*1 + i] - temp[8*6 + i];
|
|
123 tmp2= temp[8*2 + i] + temp[8*5 + i];
|
|
124 tmp5= temp[8*2 + i] - temp[8*5 + i];
|
|
125 tmp3= temp[8*3 + i] + temp[8*4 + i];
|
|
126 tmp4= temp[8*3 + i] - temp[8*4 + i];
|
|
127
|
|
128 tmp10= tmp0 + tmp3;
|
|
129 tmp13= tmp0 - tmp3;
|
|
130 tmp11= tmp1 + tmp2;
|
|
131 tmp12= tmp1 - tmp2;
|
|
132
|
|
133 data[8*0 + i]= lrint(postscale[8*0 + i] * (tmp10 + tmp11));
|
|
134 data[8*4 + i]= lrint(postscale[8*4 + i] * (tmp10 - tmp11));
|
|
135
|
|
136 z1= (tmp12 + tmp13)* A1;
|
|
137 data[8*2 + i]= lrint(postscale[8*2 + i] * (tmp13 + z1));
|
|
138 data[8*6 + i]= lrint(postscale[8*6 + i] * (tmp13 - z1));
|
|
139
|
|
140 tmp10= tmp4 + tmp5;
|
|
141 tmp11= tmp5 + tmp6;
|
|
142 tmp12= tmp6 + tmp7;
|
|
143
|
|
144 z5= (tmp10 - tmp12) * A5;
|
|
145 z2= tmp10*A2 + z5;
|
|
146 z4= tmp12*A4 + z5;
|
|
147 z3= tmp11*A1;
|
|
148
|
|
149 z11= tmp7 + z3;
|
|
150 z13= tmp7 - z3;
|
|
151
|
|
152 data[8*5 + i]= lrint(postscale[8*5 + i] * (z13 + z2));
|
|
153 data[8*3 + i]= lrint(postscale[8*3 + i] * (z13 - z2));
|
|
154 data[8*1 + i]= lrint(postscale[8*1 + i] * (z11 + z4));
|
|
155 data[8*7 + i]= lrint(postscale[8*7 + i] * (z11 - z4));
|
|
156 }
|
|
157 }
|