1 /*
2 * Copyright (c) 2002-2007, Communications and Remote Sensing Laboratory, Universite catholique de Louvain (UCL), Belgium
3 * Copyright (c) 2002-2007, Professor Benoit Macq
4 * Copyright (c) 2001-2003, David Janssens
5 * Copyright (c) 2002-2003, Yannick Verschueren
6 * Copyright (c) 2003-2007, Francois-Olivier Devaux and Antonin Descampe
7 * Copyright (c) 2005, Herve Drolon, FreeImage Team
8 * Copyright (c) 2008;2011-2012, Centre National d'Etudes Spatiales (CNES), France
9 * Copyright (c) 2012, CS Systemes d'Information, France
10 * All rights reserved.
11 *
12 * Redistribution and use in source and binary forms, with or without
13 * modification, are permitted provided that the following conditions
14 * are met:
15 * 1. Redistributions of source code must retain the above copyright
16 * notice, this list of conditions and the following disclaimer.
17 * 2. Redistributions in binary form must reproduce the above copyright
18 * notice, this list of conditions and the following disclaimer in the
19 * documentation and/or other materials provided with the distribution.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS `AS IS'
22 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
25 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
26 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
27 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
28 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
29 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
30 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
31 * POSSIBILITY OF SUCH DAMAGE.
32 */
33
34 #ifdef __SSE__
35 #include <xmmintrin.h>
36 #endif
37
38 #include "opj_includes.h"
39
40 /* <summary> */
41 /* This table contains the norms of the basis function of the reversible MCT. */
42 /* </summary> */
43 static const OPJ_FLOAT64 opj_mct_norms[3] = { 1.732, .8292, .8292 };
44
45 /* <summary> */
46 /* This table contains the norms of the basis function of the irreversible MCT. */
47 /* </summary> */
48 static const OPJ_FLOAT64 opj_mct_norms_real[3] = { 1.732, 1.805, 1.573 };
49
opj_mct_get_mct_norms()50 const OPJ_FLOAT64 * opj_mct_get_mct_norms ()
51 {
52 return opj_mct_norms;
53 }
54
opj_mct_get_mct_norms_real()55 const OPJ_FLOAT64 * opj_mct_get_mct_norms_real ()
56 {
57 return opj_mct_norms_real;
58 }
59
60 /* <summary> */
61 /* Foward reversible MCT. */
62 /* </summary> */
opj_mct_encode(OPJ_INT32 * restrict c0,OPJ_INT32 * restrict c1,OPJ_INT32 * restrict c2,OPJ_UINT32 n)63 void opj_mct_encode(
64 OPJ_INT32* restrict c0,
65 OPJ_INT32* restrict c1,
66 OPJ_INT32* restrict c2,
67 OPJ_UINT32 n)
68 {
69 OPJ_UINT32 i;
70 for(i = 0; i < n; ++i) {
71 OPJ_INT32 r = c0[i];
72 OPJ_INT32 g = c1[i];
73 OPJ_INT32 b = c2[i];
74 OPJ_INT32 y = (r + (g * 2) + b) >> 2;
75 OPJ_INT32 u = b - g;
76 OPJ_INT32 v = r - g;
77 c0[i] = y;
78 c1[i] = u;
79 c2[i] = v;
80 }
81 }
82
83 /* <summary> */
84 /* Inverse reversible MCT. */
85 /* </summary> */
opj_mct_decode(OPJ_INT32 * restrict c0,OPJ_INT32 * restrict c1,OPJ_INT32 * restrict c2,OPJ_UINT32 n)86 void opj_mct_decode(
87 OPJ_INT32* restrict c0,
88 OPJ_INT32* restrict c1,
89 OPJ_INT32* restrict c2,
90 OPJ_UINT32 n)
91 {
92 OPJ_UINT32 i;
93 for (i = 0; i < n; ++i) {
94 OPJ_INT32 y = c0[i];
95 OPJ_INT32 u = c1[i];
96 OPJ_INT32 v = c2[i];
97 OPJ_INT32 g = y - ((u + v) >> 2);
98 OPJ_INT32 r = v + g;
99 OPJ_INT32 b = u + g;
100 c0[i] = r;
101 c1[i] = g;
102 c2[i] = b;
103 }
104 }
105
106 /* <summary> */
107 /* Get norm of basis function of reversible MCT. */
108 /* </summary> */
opj_mct_getnorm(OPJ_UINT32 compno)109 OPJ_FLOAT64 opj_mct_getnorm(OPJ_UINT32 compno) {
110 return opj_mct_norms[compno];
111 }
112
113 /* <summary> */
114 /* Foward irreversible MCT. */
115 /* </summary> */
opj_mct_encode_real(OPJ_INT32 * restrict c0,OPJ_INT32 * restrict c1,OPJ_INT32 * restrict c2,OPJ_UINT32 n)116 void opj_mct_encode_real(
117 OPJ_INT32* restrict c0,
118 OPJ_INT32* restrict c1,
119 OPJ_INT32* restrict c2,
120 OPJ_UINT32 n)
121 {
122 OPJ_UINT32 i;
123 for(i = 0; i < n; ++i) {
124 OPJ_INT32 r = c0[i];
125 OPJ_INT32 g = c1[i];
126 OPJ_INT32 b = c2[i];
127 OPJ_INT32 y = opj_int_fix_mul(r, 2449) + opj_int_fix_mul(g, 4809) + opj_int_fix_mul(b, 934);
128 OPJ_INT32 u = -opj_int_fix_mul(r, 1382) - opj_int_fix_mul(g, 2714) + opj_int_fix_mul(b, 4096);
129 OPJ_INT32 v = opj_int_fix_mul(r, 4096) - opj_int_fix_mul(g, 3430) - opj_int_fix_mul(b, 666);
130 c0[i] = y;
131 c1[i] = u;
132 c2[i] = v;
133 }
134 }
135
136 /* <summary> */
137 /* Inverse irreversible MCT. */
138 /* </summary> */
opj_mct_decode_real(OPJ_FLOAT32 * restrict c0,OPJ_FLOAT32 * restrict c1,OPJ_FLOAT32 * restrict c2,OPJ_UINT32 n)139 void opj_mct_decode_real(
140 OPJ_FLOAT32* restrict c0,
141 OPJ_FLOAT32* restrict c1,
142 OPJ_FLOAT32* restrict c2,
143 OPJ_UINT32 n)
144 {
145 OPJ_UINT32 i;
146 #ifdef __SSE__
147 __m128 vrv, vgu, vgv, vbu;
148 vrv = _mm_set1_ps(1.402f);
149 vgu = _mm_set1_ps(0.34413f);
150 vgv = _mm_set1_ps(0.71414f);
151 vbu = _mm_set1_ps(1.772f);
152 for (i = 0; i < (n >> 3); ++i) {
153 __m128 vy, vu, vv;
154 __m128 vr, vg, vb;
155
156 vy = _mm_load_ps(c0);
157 vu = _mm_load_ps(c1);
158 vv = _mm_load_ps(c2);
159 vr = _mm_add_ps(vy, _mm_mul_ps(vv, vrv));
160 vg = _mm_sub_ps(_mm_sub_ps(vy, _mm_mul_ps(vu, vgu)), _mm_mul_ps(vv, vgv));
161 vb = _mm_add_ps(vy, _mm_mul_ps(vu, vbu));
162 _mm_store_ps(c0, vr);
163 _mm_store_ps(c1, vg);
164 _mm_store_ps(c2, vb);
165 c0 += 4;
166 c1 += 4;
167 c2 += 4;
168
169 vy = _mm_load_ps(c0);
170 vu = _mm_load_ps(c1);
171 vv = _mm_load_ps(c2);
172 vr = _mm_add_ps(vy, _mm_mul_ps(vv, vrv));
173 vg = _mm_sub_ps(_mm_sub_ps(vy, _mm_mul_ps(vu, vgu)), _mm_mul_ps(vv, vgv));
174 vb = _mm_add_ps(vy, _mm_mul_ps(vu, vbu));
175 _mm_store_ps(c0, vr);
176 _mm_store_ps(c1, vg);
177 _mm_store_ps(c2, vb);
178 c0 += 4;
179 c1 += 4;
180 c2 += 4;
181 }
182 n &= 7;
183 #endif
184 for(i = 0; i < n; ++i) {
185 OPJ_FLOAT32 y = c0[i];
186 OPJ_FLOAT32 u = c1[i];
187 OPJ_FLOAT32 v = c2[i];
188 OPJ_FLOAT32 r = y + (v * 1.402f);
189 OPJ_FLOAT32 g = y - (u * 0.34413f) - (v * (0.71414f));
190 OPJ_FLOAT32 b = y + (u * 1.772f);
191 c0[i] = r;
192 c1[i] = g;
193 c2[i] = b;
194 }
195 }
196
197 /* <summary> */
198 /* Get norm of basis function of irreversible MCT. */
199 /* </summary> */
opj_mct_getnorm_real(OPJ_UINT32 compno)200 OPJ_FLOAT64 opj_mct_getnorm_real(OPJ_UINT32 compno) {
201 return opj_mct_norms_real[compno];
202 }
203
204
opj_mct_encode_custom(OPJ_BYTE * pCodingdata,OPJ_UINT32 n,OPJ_BYTE ** pData,OPJ_UINT32 pNbComp,OPJ_UINT32 isSigned)205 OPJ_BOOL opj_mct_encode_custom(
206 OPJ_BYTE * pCodingdata,
207 OPJ_UINT32 n,
208 OPJ_BYTE ** pData,
209 OPJ_UINT32 pNbComp,
210 OPJ_UINT32 isSigned)
211 {
212 OPJ_FLOAT32 * lMct = (OPJ_FLOAT32 *) pCodingdata;
213 OPJ_UINT32 i;
214 OPJ_UINT32 j;
215 OPJ_UINT32 k;
216 OPJ_UINT32 lNbMatCoeff = pNbComp * pNbComp;
217 OPJ_INT32 * lCurrentData = 00;
218 OPJ_INT32 * lCurrentMatrix = 00;
219 OPJ_INT32 ** lData = (OPJ_INT32 **) pData;
220 OPJ_UINT32 lMultiplicator = 1 << 13;
221 OPJ_INT32 * lMctPtr;
222
223 OPJ_ARG_NOT_USED(isSigned);
224
225 lCurrentData = (OPJ_INT32 *) opj_malloc((pNbComp + lNbMatCoeff) * sizeof(OPJ_INT32));
226 if (! lCurrentData) {
227 return OPJ_FALSE;
228 }
229
230 lCurrentMatrix = lCurrentData + pNbComp;
231
232 for (i =0;i<lNbMatCoeff;++i) {
233 lCurrentMatrix[i] = (OPJ_INT32) (*(lMct++) * (OPJ_FLOAT32)lMultiplicator);
234 }
235
236 for (i = 0; i < n; ++i) {
237 lMctPtr = lCurrentMatrix;
238 for (j=0;j<pNbComp;++j) {
239 lCurrentData[j] = (*(lData[j]));
240 }
241
242 for (j=0;j<pNbComp;++j) {
243 *(lData[j]) = 0;
244 for (k=0;k<pNbComp;++k) {
245 *(lData[j]) += opj_int_fix_mul(*lMctPtr, lCurrentData[k]);
246 ++lMctPtr;
247 }
248
249 ++lData[j];
250 }
251 }
252
253 opj_free(lCurrentData);
254
255 return OPJ_TRUE;
256 }
257
opj_mct_decode_custom(OPJ_BYTE * pDecodingData,OPJ_UINT32 n,OPJ_BYTE ** pData,OPJ_UINT32 pNbComp,OPJ_UINT32 isSigned)258 OPJ_BOOL opj_mct_decode_custom(
259 OPJ_BYTE * pDecodingData,
260 OPJ_UINT32 n,
261 OPJ_BYTE ** pData,
262 OPJ_UINT32 pNbComp,
263 OPJ_UINT32 isSigned)
264 {
265 OPJ_FLOAT32 * lMct;
266 OPJ_UINT32 i;
267 OPJ_UINT32 j;
268 OPJ_UINT32 k;
269
270 OPJ_FLOAT32 * lCurrentData = 00;
271 OPJ_FLOAT32 * lCurrentResult = 00;
272 OPJ_FLOAT32 ** lData = (OPJ_FLOAT32 **) pData;
273
274 OPJ_ARG_NOT_USED(isSigned);
275
276 lCurrentData = (OPJ_FLOAT32 *) opj_malloc (2 * pNbComp * sizeof(OPJ_FLOAT32));
277 if (! lCurrentData) {
278 return OPJ_FALSE;
279 }
280 lCurrentResult = lCurrentData + pNbComp;
281
282 for (i = 0; i < n; ++i) {
283 lMct = (OPJ_FLOAT32 *) pDecodingData;
284 for (j=0;j<pNbComp;++j) {
285 lCurrentData[j] = (OPJ_FLOAT32) (*(lData[j]));
286 }
287 for (j=0;j<pNbComp;++j) {
288 lCurrentResult[j] = 0;
289 for (k=0;k<pNbComp;++k) {
290 lCurrentResult[j] += *(lMct++) * lCurrentData[k];
291 }
292 *(lData[j]++) = (OPJ_FLOAT32) (lCurrentResult[j]);
293 }
294 }
295 opj_free(lCurrentData);
296 return OPJ_TRUE;
297 }
298
opj_calculate_norms(OPJ_FLOAT64 * pNorms,OPJ_UINT32 pNbComps,OPJ_FLOAT32 * pMatrix)299 void opj_calculate_norms( OPJ_FLOAT64 * pNorms,
300 OPJ_UINT32 pNbComps,
301 OPJ_FLOAT32 * pMatrix)
302 {
303 OPJ_UINT32 i,j,lIndex;
304 OPJ_FLOAT32 lCurrentValue;
305 OPJ_FLOAT64 * lNorms = (OPJ_FLOAT64 *) pNorms;
306 OPJ_FLOAT32 * lMatrix = (OPJ_FLOAT32 *) pMatrix;
307
308 for (i=0;i<pNbComps;++i) {
309 lNorms[i] = 0;
310 lIndex = i;
311
312 for (j=0;j<pNbComps;++j) {
313 lCurrentValue = lMatrix[lIndex];
314 lIndex += pNbComps;
315 lNorms[i] += lCurrentValue * lCurrentValue;
316 }
317 lNorms[i] = sqrt(lNorms[i]);
318 }
319 }
320