1 /* 2 * Copyright (c) 1997, 2003, Oracle and/or its affiliates. All rights reserved. 3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. 4 * 5 * This code is free software; you can redistribute it and/or modify it 6 * under the terms of the GNU General Public License version 2 only, as 7 * published by the Free Software Foundation. Oracle designates this 8 * particular file as subject to the "Classpath" exception as provided 9 * by Oracle in the LICENSE file that accompanied this code. 10 * 11 * This code is distributed in the hope that it will be useful, but WITHOUT 12 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or 13 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License 14 * version 2 for more details (a copy is included in the LICENSE file that 15 * accompanied this code). 16 * 17 * You should have received a copy of the GNU General Public License version 18 * 2 along with this work; if not, write to the Free Software Foundation, 19 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. 20 * 21 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA 22 * or visit www.oracle.com if you need additional information or have any 23 * questions. 24 */ 25 26 27 /* 28 * FUNCTION 29 * mlib_ImageAffine_u8_1ch_bl 30 * mlib_ImageAffine_u8_2ch_bl 31 * mlib_ImageAffine_u8_3ch_bl 32 * mlib_ImageAffine_u8_4ch_bl 33 * - image affine transformation with Bilinear filtering 34 * SYNOPSIS 35 * mlib_status mlib_ImageAffine_u8_?ch_bl(mlib_s32 *leftEdges, 36 * mlib_s32 *rightEdges, 37 * mlib_s32 *xStarts, 38 * mlib_s32 *yStarts, 39 * mlib_s32 *sides, 40 * mlib_u8 *dstData, 41 * mlib_u8 **lineAddr, 42 * mlib_s32 dstYStride, 43 * mlib_s32 is_affine, 44 * mlib_s32 srcYStride) 45 * 46 * ARGUMENTS 47 * leftEdges array[dstHeight] of xLeft coordinates 48 * RightEdges array[dstHeight] of xRight coordinates 49 * xStarts array[dstHeight] of xStart * 65536 coordinates 50 * yStarts array[dstHeight] of yStart * 65536 coordinates 51 * sides output array[4]. sides[0] is yStart, sides[1] is yFinish, 52 * sides[2] is dx * 65536, sides[3] is dy * 65536 53 * dstData pointer to the first pixel on (yStart - 1) line 54 * lineAddr array[srcHeight] of pointers to the first pixel on 55 * the corresponding lines 56 * dstYStride stride of destination image 57 * is_affine indicator (Affine - GridWarp) 58 * srcYStride stride of source image 59 * 60 * DESCRIPTION 61 * The functions step along the lines from xLeft to xRight and apply 62 * the bilinear filtering. 63 * 64 */ 65 66 #include "mlib_ImageAffine.h" 67 68 /***************************************************************/ 69 #define DTYPE mlib_u8 70 #define FTYPE mlib_f32 71 72 /***************************************************************/ 73 #define TTYPE mlib_f32 74 #define I2F(x) mlib_U82F32[x] 75 #define ROUND(x) ((x) + 0.5f) 76 77 #define FUN_NAME(CHAN) mlib_ImageAffine_u8_##CHAN##_bl 78 79 /***************************************************************/ 80 #ifdef __sparc /* for SPARC, using floating-point multiplies is faster */ 81 82 /***************************************************************/ 83 #define GET_POINTERS(ind) \ 84 fdx = (FTYPE)(X & MLIB_MASK) * scale; \ 85 fdy = (FTYPE)(Y & MLIB_MASK) * scale; \ 86 ySrc = MLIB_POINTER_SHIFT(Y); Y += dY; \ 87 xSrc = X >> MLIB_SHIFT; X += dX; \ 88 srcPixelPtr = MLIB_POINTER_GET(lineAddr, ySrc) + ind * xSrc; \ 89 srcPixelPtr2 = (DTYPE *)((mlib_u8 *)srcPixelPtr + srcYStride) 90 91 /***************************************************************/ 92 #define COUNT(ind) \ 93 pix0_##ind = a00_##ind + fdy * (a10_##ind - a00_##ind); \ 94 pix1_##ind = a01_##ind + fdy * (a11_##ind - a01_##ind); \ 95 res##ind = ROUND(pix0_##ind + fdx * (pix1_##ind - pix0_##ind)) 96 97 /***************************************************************/ 98 #define LOAD(ind, ind1, ind2) \ 99 a00_##ind = I2F(srcPixelPtr[ind1]); \ 100 a01_##ind = I2F(srcPixelPtr[ind2]); \ 101 a10_##ind = I2F(srcPixelPtr2[ind1]); \ 102 a11_##ind = I2F(srcPixelPtr2[ind2]) 103 104 /***************************************************************/ 105 mlib_status FUN_NAME(1ch)(mlib_affine_param *param) 106 { 107 DECLAREVAR_BL(); 108 DTYPE *dstLineEnd; 109 DTYPE *srcPixelPtr2; 110 FTYPE scale = (FTYPE) 1.0 / MLIB_PREC; 111 112 for (j = yStart; j <= yFinish; j++) { 113 FTYPE fdx, fdy; 114 TTYPE a00_0, a01_0, a10_0, a11_0; 115 FTYPE pix0_0, pix1_0, res0; 116 117 CLIP(1); 118 dstLineEnd = (DTYPE *) dstData + xRight; 119 120 GET_POINTERS(1); 121 LOAD(0, 0, 1); 122 123 #ifdef __SUNPRO_C 124 #pragma pipeloop(0) 125 #endif /* __SUNPRO_C */ 126 for (; dstPixelPtr < dstLineEnd; dstPixelPtr++) { 127 COUNT(0); 128 GET_POINTERS(1); 129 LOAD(0, 0, 1); 130 dstPixelPtr[0] = (DTYPE) res0; 131 } 132 133 COUNT(0); 134 dstPixelPtr[0] = (DTYPE) res0; 135 } 136 137 return MLIB_SUCCESS; 138 } 139 140 /***************************************************************/ 141 mlib_status FUN_NAME(2ch)(mlib_affine_param *param) 142 { 143 DECLAREVAR_BL(); 144 DTYPE *dstLineEnd; 145 DTYPE *srcPixelPtr2; 146 FTYPE scale = (FTYPE) 1.0 / MLIB_PREC; 147 148 for (j = yStart; j <= yFinish; j++) { 149 FTYPE fdx, fdy; 150 TTYPE a00_0, a01_0, a10_0, a11_0; 151 TTYPE a00_1, a01_1, a10_1, a11_1; 152 FTYPE pix0_0, pix1_0, res0; 153 FTYPE pix0_1, pix1_1, res1; 154 155 CLIP(2); 156 dstLineEnd = (DTYPE *) dstData + 2 * xRight; 157 158 GET_POINTERS(2); 159 LOAD(0, 0, 2); 160 LOAD(1, 1, 3); 161 162 #ifdef __SUNPRO_C 163 #pragma pipeloop(0) 164 #endif /* __SUNPRO_C */ 165 for (; dstPixelPtr < dstLineEnd; dstPixelPtr += 2) { 166 COUNT(0); 167 COUNT(1); 168 GET_POINTERS(2); 169 LOAD(0, 0, 2); 170 LOAD(1, 1, 3); 171 dstPixelPtr[0] = (DTYPE) res0; 172 dstPixelPtr[1] = (DTYPE) res1; 173 } 174 175 COUNT(0); 176 COUNT(1); 177 dstPixelPtr[0] = (DTYPE) res0; 178 dstPixelPtr[1] = (DTYPE) res1; 179 } 180 181 return MLIB_SUCCESS; 182 } 183 184 /***************************************************************/ 185 mlib_status FUN_NAME(3ch)(mlib_affine_param *param) 186 { 187 DECLAREVAR_BL(); 188 DTYPE *dstLineEnd; 189 DTYPE *srcPixelPtr2; 190 FTYPE scale = (FTYPE) 1.0 / MLIB_PREC; 191 192 for (j = yStart; j <= yFinish; j++) { 193 FTYPE fdx, fdy; 194 FTYPE a00_0, a01_0, a10_0, a11_0; 195 FTYPE a00_1, a01_1, a10_1, a11_1; 196 FTYPE a00_2, a01_2, a10_2, a11_2; 197 FTYPE pix0_0, pix1_0, res0; 198 FTYPE pix0_1, pix1_1, res1; 199 FTYPE pix0_2, pix1_2, res2; 200 201 CLIP(3); 202 dstLineEnd = (DTYPE *) dstData + 3 * xRight; 203 204 GET_POINTERS(3); 205 LOAD(0, 0, 3); 206 LOAD(1, 1, 4); 207 LOAD(2, 2, 5); 208 209 #ifdef __SUNPRO_C 210 #pragma pipeloop(0) 211 #endif /* __SUNPRO_C */ 212 for (; dstPixelPtr < dstLineEnd; dstPixelPtr += 3) { 213 COUNT(0); 214 COUNT(1); 215 COUNT(2); 216 GET_POINTERS(3); 217 LOAD(0, 0, 3); 218 LOAD(1, 1, 4); 219 LOAD(2, 2, 5); 220 dstPixelPtr[0] = (DTYPE) res0; 221 dstPixelPtr[1] = (DTYPE) res1; 222 dstPixelPtr[2] = (DTYPE) res2; 223 } 224 225 COUNT(0); 226 COUNT(1); 227 COUNT(2); 228 dstPixelPtr[0] = (DTYPE) res0; 229 dstPixelPtr[1] = (DTYPE) res1; 230 dstPixelPtr[2] = (DTYPE) res2; 231 } 232 233 return MLIB_SUCCESS; 234 } 235 236 /***************************************************************/ 237 mlib_status FUN_NAME(4ch)(mlib_affine_param *param) 238 { 239 DECLAREVAR_BL(); 240 DTYPE *dstLineEnd; 241 DTYPE *srcPixelPtr2; 242 FTYPE scale = (FTYPE) 1.0 / MLIB_PREC; 243 244 for (j = yStart; j <= yFinish; j++) { 245 FTYPE fdx, fdy; 246 TTYPE a00_0, a01_0, a10_0, a11_0; 247 TTYPE a00_1, a01_1, a10_1, a11_1; 248 TTYPE a00_2, a01_2, a10_2, a11_2; 249 TTYPE a00_3, a01_3, a10_3, a11_3; 250 FTYPE pix0_0, pix1_0, res0; 251 FTYPE pix0_1, pix1_1, res1; 252 FTYPE pix0_2, pix1_2, res2; 253 FTYPE pix0_3, pix1_3, res3; 254 255 CLIP(4); 256 dstLineEnd = (DTYPE *) dstData + 4 * xRight; 257 258 GET_POINTERS(4); 259 LOAD(0, 0, 4); 260 LOAD(1, 1, 5); 261 LOAD(2, 2, 6); 262 LOAD(3, 3, 7); 263 264 #ifdef __SUNPRO_C 265 #pragma pipeloop(0) 266 #endif /* __SUNPRO_C */ 267 for (; dstPixelPtr < dstLineEnd; dstPixelPtr += 4) { 268 COUNT(0); 269 COUNT(1); 270 COUNT(2); 271 COUNT(3); 272 GET_POINTERS(4); 273 LOAD(0, 0, 4); 274 LOAD(1, 1, 5); 275 LOAD(2, 2, 6); 276 LOAD(3, 3, 7); 277 dstPixelPtr[0] = (DTYPE) res0; 278 dstPixelPtr[1] = (DTYPE) res1; 279 dstPixelPtr[2] = (DTYPE) res2; 280 dstPixelPtr[3] = (DTYPE) res3; 281 } 282 283 COUNT(0); 284 COUNT(1); 285 COUNT(2); 286 COUNT(3); 287 dstPixelPtr[0] = (DTYPE) res0; 288 dstPixelPtr[1] = (DTYPE) res1; 289 dstPixelPtr[2] = (DTYPE) res2; 290 dstPixelPtr[3] = (DTYPE) res3; 291 } 292 293 return MLIB_SUCCESS; 294 } 295 296 #else /* for x86, using integer multiplies is faster */ 297 298 /* for SHORT/USHORT decrease MLIB_SHIFT due to 299 * overflow in multiplies like fdy * (a10 - a00) 300 */ 301 /* 302 #undef MLIB_SHIFT 303 #define MLIB_SHIFT 15 304 */ 305 306 #define MLIB_ROUND (1 << (MLIB_SHIFT - 1)) 307 308 /***************************************************************/ 309 #define GET_POINTERS(ind) \ 310 fdx = X & MLIB_MASK; \ 311 fdy = Y & MLIB_MASK; \ 312 ySrc = MLIB_POINTER_SHIFT(Y); \ 313 xSrc = X >> MLIB_SHIFT; \ 314 srcPixelPtr = MLIB_POINTER_GET(lineAddr, ySrc) + ind * xSrc; \ 315 srcPixelPtr2 = (DTYPE *)((mlib_u8 *)srcPixelPtr + srcYStride); \ 316 X += dX; \ 317 Y += dY 318 319 /***************************************************************/ 320 #define COUNT(ind) \ 321 pix0_##ind = a00_##ind + ((fdy * (a10_##ind - a00_##ind) + MLIB_ROUND) >> MLIB_SHIFT); \ 322 pix1_##ind = a01_##ind + ((fdy * (a11_##ind - a01_##ind) + MLIB_ROUND) >> MLIB_SHIFT); \ 323 res##ind = pix0_##ind + ((fdx * (pix1_##ind - pix0_##ind) + MLIB_ROUND) >> MLIB_SHIFT) 324 325 /***************************************************************/ 326 #define LOAD(ind, ind1, ind2) \ 327 a00_##ind = srcPixelPtr[ind1]; \ 328 a01_##ind = srcPixelPtr[ind2]; \ 329 a10_##ind = srcPixelPtr2[ind1]; \ 330 a11_##ind = srcPixelPtr2[ind2] 331 332 /***************************************************************/ 333 mlib_status FUN_NAME(1ch)(mlib_affine_param *param) 334 { 335 DECLAREVAR_BL(); 336 DTYPE *dstLineEnd; 337 DTYPE *srcPixelPtr2; 338 339 #if MLIB_SHIFT == 15 340 dX = (dX + 1) >> 1; 341 dY = (dY + 1) >> 1; 342 #endif /* MLIB_SHIFT == 15 */ 343 344 for (j = yStart; j <= yFinish; j++) { 345 mlib_s32 fdx, fdy; 346 mlib_s32 a00_0, a01_0, a10_0, a11_0; 347 mlib_s32 pix0_0, pix1_0, res0; 348 349 CLIP(1); 350 dstLineEnd = (DTYPE *) dstData + xRight; 351 #if MLIB_SHIFT == 15 352 X = X >> 1; 353 Y = Y >> 1; 354 #endif /* MLIB_SHIFT == 15 */ 355 356 GET_POINTERS(1); 357 LOAD(0, 0, 1); 358 359 #ifdef __SUNPRO_C 360 #pragma pipeloop(0) 361 #endif /* __SUNPRO_C */ 362 for (; dstPixelPtr < dstLineEnd; dstPixelPtr++) { 363 COUNT(0); 364 GET_POINTERS(1); 365 LOAD(0, 0, 1); 366 dstPixelPtr[0] = (DTYPE) res0; 367 } 368 369 COUNT(0); 370 dstPixelPtr[0] = (DTYPE) res0; 371 } 372 373 return MLIB_SUCCESS; 374 } 375 376 /***************************************************************/ 377 mlib_status FUN_NAME(2ch)(mlib_affine_param *param) 378 { 379 DECLAREVAR_BL(); 380 DTYPE *dstLineEnd; 381 DTYPE *srcPixelPtr2; 382 383 #if MLIB_SHIFT == 15 384 dX = (dX + 1) >> 1; 385 dY = (dY + 1) >> 1; 386 #endif /* MLIB_SHIFT == 15 */ 387 388 for (j = yStart; j <= yFinish; j++) { 389 mlib_s32 fdx, fdy; 390 mlib_s32 a00_0, a01_0, a10_0, a11_0; 391 mlib_s32 a00_1, a01_1, a10_1, a11_1; 392 mlib_s32 pix0_0, pix1_0, res0; 393 mlib_s32 pix0_1, pix1_1, res1; 394 395 CLIP(2); 396 dstLineEnd = (DTYPE *) dstData + 2 * xRight; 397 #if MLIB_SHIFT == 15 398 X = X >> 1; 399 Y = Y >> 1; 400 #endif /* MLIB_SHIFT == 15 */ 401 402 GET_POINTERS(2); 403 LOAD(0, 0, 2); 404 LOAD(1, 1, 3); 405 406 #ifdef __SUNPRO_C 407 #pragma pipeloop(0) 408 #endif /* __SUNPRO_C */ 409 for (; dstPixelPtr < dstLineEnd; dstPixelPtr += 2) { 410 COUNT(0); 411 COUNT(1); 412 GET_POINTERS(2); 413 LOAD(0, 0, 2); 414 LOAD(1, 1, 3); 415 dstPixelPtr[0] = (DTYPE) res0; 416 dstPixelPtr[1] = (DTYPE) res1; 417 } 418 419 COUNT(0); 420 COUNT(1); 421 dstPixelPtr[0] = (DTYPE) res0; 422 dstPixelPtr[1] = (DTYPE) res1; 423 } 424 425 return MLIB_SUCCESS; 426 } 427 428 /***************************************************************/ 429 mlib_status FUN_NAME(3ch)(mlib_affine_param *param) 430 { 431 DECLAREVAR_BL(); 432 DTYPE *dstLineEnd; 433 DTYPE *srcPixelPtr2; 434 435 #if MLIB_SHIFT == 15 436 dX = (dX + 1) >> 1; 437 dY = (dY + 1) >> 1; 438 #endif /* MLIB_SHIFT == 15 */ 439 440 for (j = yStart; j <= yFinish; j++) { 441 mlib_s32 fdx, fdy; 442 mlib_s32 a00_0, a01_0, a10_0, a11_0; 443 mlib_s32 a00_1, a01_1, a10_1, a11_1; 444 mlib_s32 a00_2, a01_2, a10_2, a11_2; 445 mlib_s32 pix0_0, pix1_0, res0; 446 mlib_s32 pix0_1, pix1_1, res1; 447 mlib_s32 pix0_2, pix1_2, res2; 448 449 CLIP(3); 450 dstLineEnd = (DTYPE *) dstData + 3 * xRight; 451 #if MLIB_SHIFT == 15 452 X = X >> 1; 453 Y = Y >> 1; 454 #endif /* MLIB_SHIFT == 15 */ 455 456 GET_POINTERS(3); 457 LOAD(0, 0, 3); 458 LOAD(1, 1, 4); 459 LOAD(2, 2, 5); 460 461 #ifdef __SUNPRO_C 462 #pragma pipeloop(0) 463 #endif /* __SUNPRO_C */ 464 for (; dstPixelPtr < dstLineEnd; dstPixelPtr += 3) { 465 COUNT(0); 466 COUNT(1); 467 COUNT(2); 468 GET_POINTERS(3); 469 LOAD(0, 0, 3); 470 LOAD(1, 1, 4); 471 LOAD(2, 2, 5); 472 dstPixelPtr[0] = (DTYPE) res0; 473 dstPixelPtr[1] = (DTYPE) res1; 474 dstPixelPtr[2] = (DTYPE) res2; 475 } 476 477 COUNT(0); 478 COUNT(1); 479 COUNT(2); 480 dstPixelPtr[0] = (DTYPE) res0; 481 dstPixelPtr[1] = (DTYPE) res1; 482 dstPixelPtr[2] = (DTYPE) res2; 483 } 484 485 return MLIB_SUCCESS; 486 } 487 488 /***************************************************************/ 489 mlib_status FUN_NAME(4ch)(mlib_affine_param *param) 490 { 491 DECLAREVAR_BL(); 492 DTYPE *dstLineEnd; 493 DTYPE *srcPixelPtr2; 494 495 #if MLIB_SHIFT == 15 496 dX = (dX + 1) >> 1; 497 dY = (dY + 1) >> 1; 498 #endif /* MLIB_SHIFT == 15 */ 499 500 for (j = yStart; j <= yFinish; j++) { 501 mlib_s32 fdx, fdy; 502 mlib_s32 a00_0, a01_0, a10_0, a11_0; 503 mlib_s32 a00_1, a01_1, a10_1, a11_1; 504 mlib_s32 a00_2, a01_2, a10_2, a11_2; 505 mlib_s32 a00_3, a01_3, a10_3, a11_3; 506 mlib_s32 pix0_0, pix1_0, res0; 507 mlib_s32 pix0_1, pix1_1, res1; 508 mlib_s32 pix0_2, pix1_2, res2; 509 mlib_s32 pix0_3, pix1_3, res3; 510 511 CLIP(4); 512 dstLineEnd = (DTYPE *) dstData + 4 * xRight; 513 #if MLIB_SHIFT == 15 514 X = X >> 1; 515 Y = Y >> 1; 516 #endif /* MLIB_SHIFT == 15 */ 517 518 GET_POINTERS(4); 519 LOAD(0, 0, 4); 520 LOAD(1, 1, 5); 521 LOAD(2, 2, 6); 522 LOAD(3, 3, 7); 523 524 #ifdef __SUNPRO_C 525 #pragma pipeloop(0) 526 #endif /* __SUNPRO_C */ 527 for (; dstPixelPtr < dstLineEnd; dstPixelPtr += 4) { 528 COUNT(0); 529 COUNT(1); 530 COUNT(2); 531 COUNT(3); 532 GET_POINTERS(4); 533 LOAD(0, 0, 4); 534 LOAD(1, 1, 5); 535 LOAD(2, 2, 6); 536 LOAD(3, 3, 7); 537 dstPixelPtr[0] = (DTYPE) res0; 538 dstPixelPtr[1] = (DTYPE) res1; 539 dstPixelPtr[2] = (DTYPE) res2; 540 dstPixelPtr[3] = (DTYPE) res3; 541 } 542 543 COUNT(0); 544 COUNT(1); 545 COUNT(2); 546 COUNT(3); 547 dstPixelPtr[0] = (DTYPE) res0; 548 dstPixelPtr[1] = (DTYPE) res1; 549 dstPixelPtr[2] = (DTYPE) res2; 550 dstPixelPtr[3] = (DTYPE) res3; 551 } 552 553 return MLIB_SUCCESS; 554 } 555 556 #endif /* __sparc ( for SPARC, using floating-point multiplies is faster ) */ 557 558 /***************************************************************/