Mercurial > hgrepos > Python2 > PyMuPDF
comparison mupdf-source/source/fitz/skew.c @ 2:b50eed0cc0ef upstream
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
The directory name has changed: no version number in the expanded directory now.
| author | Franz Glasner <fzglas.hg@dom66.de> |
|---|---|
| date | Mon, 15 Sep 2025 11:43:07 +0200 |
| parents | |
| children |
comparison
equal
deleted
inserted
replaced
| 1:1d09e1dec1d9 | 2:b50eed0cc0ef |
|---|---|
| 1 // Copyright (C) 2004-2021 Artifex Software, Inc. | |
| 2 // | |
| 3 // This file is part of MuPDF. | |
| 4 // | |
| 5 // MuPDF is free software: you can redistribute it and/or modify it under the | |
| 6 // terms of the GNU Affero General Public License as published by the Free | |
| 7 // Software Foundation, either version 3 of the License, or (at your option) | |
| 8 // any later version. | |
| 9 // | |
| 10 // MuPDF is distributed in the hope that it will be useful, but WITHOUT ANY | |
| 11 // WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS | |
| 12 // FOR A PARTICULAR PURPOSE. See the GNU Affero General Public License for more | |
| 13 // details. | |
| 14 // | |
| 15 // You should have received a copy of the GNU Affero General Public License | |
| 16 // along with MuPDF. If not, see <https://www.gnu.org/licenses/agpl-3.0.en.html> | |
| 17 // | |
| 18 // Alternative licensing terms are available from the licensor. | |
| 19 // For commercial licensing, see <https://www.artifex.com/> or contact | |
| 20 // Artifex Software, Inc., 39 Mesa Street, Suite 108A, San Francisco, | |
| 21 // CA 94129, USA, for further information. | |
| 22 | |
| 23 #include "mupdf/fitz.h" | |
| 24 | |
| 25 #include <math.h> | |
| 26 #include <assert.h> | |
| 27 #include <limits.h> | |
| 28 | |
| 29 #if ARCH_HAS_SSE | |
| 30 #include <emmintrin.h> | |
| 31 #include <smmintrin.h> | |
| 32 #endif | |
| 33 | |
| 34 /* Uncomment the following to enable Debugging printfs. */ | |
| 35 /* #define DEBUG_PRINT_WORKING */ | |
| 36 | |
| 37 enum { | |
| 38 NUM_SKEW_COLS = 4, | |
| 39 SKEW_COL_OFFSET = 96, | |
| 40 SKEW_COL_WIDTH = 96, | |
| 41 SKEW_MAX_DIFF = 16 | |
| 42 }; | |
| 43 | |
| 44 typedef struct | |
| 45 { | |
| 46 void *src; | |
| 47 uint32_t src_w; | |
| 48 uint32_t src_h; | |
| 49 uint32_t y; | |
| 50 uint32_t offsets[NUM_SKEW_COLS*2]; | |
| 51 uint16_t *tables[NUM_SKEW_COLS*2]; | |
| 52 uint32_t corr_height; | |
| 53 } fz_skew; | |
| 54 | |
| 55 /* Finalise a rescaler instance. */ | |
| 56 static void | |
| 57 fz_drop_skew(fz_context *ctx, fz_skew *skew) | |
| 58 { | |
| 59 if (skew == NULL) | |
| 60 return; | |
| 61 | |
| 62 fz_free(ctx, skew->tables[0]); | |
| 63 fz_free(ctx, skew); | |
| 64 } | |
| 65 | |
| 66 /* Initialise a skew instance. */ | |
| 67 static fz_skew * | |
| 68 fz_new_skew(fz_context *ctx, | |
| 69 unsigned int src_w, | |
| 70 unsigned int src_h) | |
| 71 { | |
| 72 fz_skew *skew = fz_malloc_struct(ctx, fz_skew); | |
| 73 int i; | |
| 74 | |
| 75 skew->src_w = src_w; | |
| 76 skew->src_h = src_h; | |
| 77 skew->y = 0; | |
| 78 skew->corr_height = src_h-100; /* FIXME */ | |
| 79 fz_try(ctx) | |
| 80 skew->tables[0] = (uint16_t *)fz_malloc(ctx, skew->src_h * sizeof(uint16_t) * NUM_SKEW_COLS * 2); | |
| 81 fz_catch(ctx) | |
| 82 { | |
| 83 fz_drop_skew(ctx, skew); | |
| 84 fz_rethrow(ctx); | |
| 85 } | |
| 86 | |
| 87 for (i = 1; i < NUM_SKEW_COLS * 2; i++) | |
| 88 skew->tables[i] = skew->tables[0] + i*skew->src_h; | |
| 89 | |
| 90 for (i = 0; i < NUM_SKEW_COLS; i++) | |
| 91 { | |
| 92 int offset = skew->src_w * (i+1) / (NUM_SKEW_COLS+1) - SKEW_COL_WIDTH/2; | |
| 93 skew->offsets[2*i ] = offset - SKEW_COL_OFFSET; | |
| 94 skew->offsets[2*i+1] = offset + SKEW_COL_OFFSET; | |
| 95 } | |
| 96 | |
| 97 return skew; | |
| 98 } | |
| 99 | |
| 100 static int sum_c(const uint8_t *data) | |
| 101 { | |
| 102 int i; | |
| 103 uint32_t sum = 0; | |
| 104 for (i = SKEW_COL_WIDTH; i > 0; i--) | |
| 105 sum += *data++; | |
| 106 return sum; | |
| 107 } | |
| 108 | |
| 109 #if ARCH_HAS_SSE | |
| 110 static inline int sum_sse(const uint8_t *data) | |
| 111 { | |
| 112 __m128i mm0, mm1, mm2, mm3, mm4, mm5; | |
| 113 __m128i mm_zero = _mm_set1_epi32(0); | |
| 114 | |
| 115 mm0 = _mm_loadu_si128((const __m128i *)(data )); // mm0 = ppoonnmmllkkjjiihhggffeeddccbbaa | |
| 116 mm1 = _mm_loadu_si128((const __m128i *)(data+16)); // mm1 = ppoonnmmllkkjjiihhggffeeddccbbaa | |
| 117 mm2 = _mm_loadu_si128((const __m128i *)(data+32)); // mm2 = ppoonnmmllkkjjiihhggffeeddccbbaa | |
| 118 mm3 = _mm_loadu_si128((const __m128i *)(data+48)); // mm3 = ppoonnmmllkkjjiihhggffeeddccbbaa | |
| 119 mm4 = _mm_loadu_si128((const __m128i *)(data+64)); // mm4 = ppoonnmmllkkjjiihhggffeeddccbbaa | |
| 120 mm5 = _mm_loadu_si128((const __m128i *)(data+80)); // mm5 = ppoonnmmllkkjjiihhggffeeddccbbaa | |
| 121 | |
| 122 mm0 = _mm_sad_epu8(mm0, mm_zero); // Max value in each half is 8*255 | |
| 123 mm1 = _mm_sad_epu8(mm1, mm_zero); | |
| 124 mm2 = _mm_sad_epu8(mm2, mm_zero); | |
| 125 mm3 = _mm_sad_epu8(mm3, mm_zero); | |
| 126 mm4 = _mm_sad_epu8(mm4, mm_zero); | |
| 127 mm5 = _mm_sad_epu8(mm5, mm_zero); | |
| 128 | |
| 129 mm0 = _mm_add_epi64(mm0, mm1); // Max value in each half is 2*8*255 | |
| 130 mm2 = _mm_add_epi64(mm2, mm3); // Max value in each half is 2*8*255 | |
| 131 mm4 = _mm_add_epi64(mm4, mm5); // Max value in each half is 2*8*255 | |
| 132 mm0 = _mm_add_epi64(mm0, mm2); // Max value in each half is 4*8*255 | |
| 133 mm0 = _mm_add_epi64(mm0, mm4); // Max value in each half is 6*8*255 | |
| 134 | |
| 135 mm0 = _mm_shuffle_epi32(mm0, (2<<2)+0); // Max value in each half is 10*8*255 | |
| 136 mm0 = _mm_hadd_epi32(mm0, mm0); // Max value in bottom bits is 20*8*255 - still fits in an unsigned 16bit word. | |
| 137 | |
| 138 return _mm_extract_epi16(mm0, 0); | |
| 139 } | |
| 140 #endif | |
| 141 | |
| 142 /* Process data: */ | |
| 143 static void | |
| 144 fz_skew_process(fz_context *ctx, fz_skew *skew, const uint8_t *input) | |
| 145 { | |
| 146 int i; | |
| 147 int off = skew->y++; | |
| 148 | |
| 149 #if ARCH_HAS_SSE | |
| 150 for (i = 0; i < NUM_SKEW_COLS * 2; i++) | |
| 151 skew->tables[i][off] = sum_sse(input + skew->offsets[i]); | |
| 152 #else | |
| 153 for (i = 0; i < NUM_SKEW_COLS * 2; i++) | |
| 154 skew->tables[i][off] = sum_c(input + skew->offsets[i]); | |
| 155 #endif | |
| 156 | |
| 157 /* Some debug code; if enabled this writes the summed value back | |
| 158 * in to give us a visual indication of where we are looking for | |
| 159 * correspondences. */ | |
| 160 #if 0 | |
| 161 for (i = 0; i < NUM_SKEW_COLS * 2; i++) { | |
| 162 int v = (skew->tables[i][off] + (SKEW_COL_WIDTH/2) ) / SKEW_COL_WIDTH; | |
| 163 memset(input + skew->offsets[i], v, SKEW_COL_WIDTH); | |
| 164 } | |
| 165 #endif | |
| 166 } | |
| 167 | |
| 168 static double | |
| 169 do_detect_skew(fz_context *ctx, fz_skew *skew) | |
| 170 { | |
| 171 int i, j, h, o, max_at; | |
| 172 int64_t max_sum, corr_at, corr_sum, avg_sum; | |
| 173 float ang; | |
| 174 int64_t avg = SKEW_COL_WIDTH * 255/2; | |
| 175 | |
| 176 if (skew == NULL) | |
| 177 return 0; | |
| 178 | |
| 179 h = skew->corr_height - 2 * SKEW_MAX_DIFF; | |
| 180 | |
| 181 corr_at = 0; | |
| 182 corr_sum = 0; | |
| 183 for (i = 0; i < NUM_SKEW_COLS; i++) | |
| 184 { | |
| 185 max_at = 9999; | |
| 186 max_sum = 0; | |
| 187 avg_sum = 0; | |
| 188 for (o = -SKEW_MAX_DIFF; o <= SKEW_MAX_DIFF; o++) | |
| 189 { | |
| 190 uint16_t *t0 = skew->tables[2*i] + SKEW_MAX_DIFF; | |
| 191 uint16_t *t1 = skew->tables[2*i+1] + SKEW_MAX_DIFF + o; | |
| 192 int64_t sum = 0; | |
| 193 for (j = 0; j < h; j++) | |
| 194 sum += ((int64_t)t0[j]-avg) * ((int64_t)t1[j]-avg); | |
| 195 if (max_sum < sum) | |
| 196 max_sum = sum, max_at = o; | |
| 197 avg_sum += sum; | |
| 198 #ifdef DEBUG_PRINT_WORKING | |
| 199 printf("col %d, offset %d -> %llx\n", i, o, sum); | |
| 200 #endif | |
| 201 } | |
| 202 avg_sum /= (SKEW_MAX_DIFF+1)*2; | |
| 203 #ifdef DEBUG_PRINT_WORKING | |
| 204 ang = (180.0 / 3.1415942) * atan(max_at / (double)(SKEW_COL_OFFSET * 2)); | |
| 205 printf("col %d max: offset %d -> %llx ang=%g\n", i, max_at, max_sum, ang); | |
| 206 #endif | |
| 207 /* Subtract the average from the maximum; we judge the significance of a | |
| 208 * match by how far it exceeds the average. max_sum becomes 'significance'. */ | |
| 209 max_sum -= avg_sum; | |
| 210 #ifdef DEBUG_PRINT_WORKING | |
| 211 printf("Significance: %llx\n", max_sum - avg_sum); | |
| 212 #endif | |
| 213 corr_at += max_at * max_sum; | |
| 214 corr_sum += max_sum; | |
| 215 } | |
| 216 ang = (180.0 / 3.1415942) * atan(corr_at / (double)(corr_sum * SKEW_COL_OFFSET * 2)); | |
| 217 | |
| 218 if (ang < -45 || ang > 45) | |
| 219 return 0; | |
| 220 | |
| 221 return ang; | |
| 222 } | |
| 223 | |
| 224 double fz_detect_skew(fz_context *ctx, fz_pixmap *pix) | |
| 225 { | |
| 226 fz_skew *skew = fz_new_skew(ctx, pix->w, pix->h); | |
| 227 int y; | |
| 228 uint8_t *ptr; | |
| 229 ptrdiff_t stride; | |
| 230 double angle; | |
| 231 fz_pixmap *pix2 = NULL; | |
| 232 | |
| 233 fz_var(pix2); | |
| 234 | |
| 235 fz_try(ctx) | |
| 236 { | |
| 237 if (pix->n != 1) | |
| 238 { | |
| 239 pix2 = fz_convert_pixmap(ctx, pix, fz_device_gray(ctx), NULL, NULL, fz_default_color_params, 0); | |
| 240 ptr = pix2->samples; | |
| 241 stride = pix2->stride; | |
| 242 } | |
| 243 else | |
| 244 { | |
| 245 ptr = pix->samples; | |
| 246 stride = pix->stride; | |
| 247 } | |
| 248 | |
| 249 for (y = pix->h; y > 0; y--) | |
| 250 { | |
| 251 fz_skew_process(ctx, skew, ptr); | |
| 252 ptr += stride; | |
| 253 } | |
| 254 | |
| 255 angle = do_detect_skew(ctx, skew); | |
| 256 } | |
| 257 fz_always(ctx) | |
| 258 { | |
| 259 fz_drop_pixmap(ctx, pix2); | |
| 260 fz_drop_skew(ctx, skew); | |
| 261 } | |
| 262 fz_catch(ctx) | |
| 263 fz_rethrow(ctx); | |
| 264 | |
| 265 return angle; | |
| 266 } |
