main(int argc,
char **argv)
{
char *dirin, *substr, *title, *fileout, *boxaafile, *boxaapath;
l_int32 ret, res, type, thresh;
l_float32 scalefactor;
BOXAA *baa;
static char mainName[] = "convertsegfilestopdf";
if (argc != 10) {
fprintf(stderr,
" Syntax: convertsegfilestopdf dirin substr res type thresh \\ \n"
" boxaafile scalefactor title fileout\n"
" where\n"
" dirin: input directory for image files\n"
" substr: Use 'allfiles' to convert all files\n"
" in the directory\n"
" res: Input resolution of each image;\n"
" assumed to all be the same\n"
" type: compression used for non-image regions:\n"
" 0: default (G4 encoding)\n"
" 1: JPEG encoding\n"
" 2: G4 encoding\n"
" 3: PNG encoding\n"
" thresh: threshold for binarization; use 0 for default\n"
" boxaafile: File of 'image' regions within each page\n"
" This contains a boxa for each page,\n"
" consisting of a set of regions\n"
" scalefactor: Use to scale down the image regions\n"
" title: Use 'none' to omit\n"
" fileout: Output pdf file\n");
return 1;
}
dirin = argv[1];
substr = argv[2];
res = atoi(argv[3]);
type = atoi(argv[4]);
thresh = atoi(argv[5]);
boxaafile = argv[6];
scalefactor = atof(argv[7]);
title = argv[8];
fileout = argv[9];
if (!strcmp(substr, "allfiles"))
substr = NULL;
if (scalefactor <= 0.0 || scalefactor > 1.0) {
L_WARNING("invalid scalefactor: setting to 1.0", mainName);
scalefactor = 1.0;
}
if (type != 1 && type != 2 && type != 3)
type = L_G4_ENCODE;
if (thresh <= 0)
thresh = 150;
if (!strcmp(title, "none"))
title = NULL;
boxaapath = genPathname(boxaafile, NULL);
if ((baa = boxaaRead(boxaapath)) == NULL) {
L_WARNING(
"boxaa file not found; converting unsegmented and unscaled",
mainName);
ret = convertFilesToPdf(dirin, substr, res, 1.0, 75, title,
fileout);
FREE(boxaapath);
return ret;
}
ret = convertSegmentedFilesToPdf(dirin, substr, res, type, thresh, baa,
75, scalefactor, title, fileout);
FREE(boxaapath);
boxaaDestroy(&baa);
return ret;
}
main(int argc,
char **argv)
{
l_int32 h;
l_float32 scalefactor;
BOX *box;
BOXA *boxa1, *boxa2;
BOXAA *baa;
PIX *pix1, *pix2, *pix3, *pix4, *pix5, *pix6, *pix7, *pix8, *pix9;
L_REGPARAMS *rp;
if (regTestSetup(argc, argv, &rp))
return 1;
lept_rmdir("segtest");
lept_mkdir("segtest");
baa = boxaaCreate(5);
/* Image region input. */
pix1 = pixRead("wet-day.jpg");
pix2 = pixScaleToSize(pix1, WIDTH, 0);
pixWrite("/tmp/segtest/0.jpg", pix2, IFF_JFIF_JPEG);
regTestCheckFile(rp, "/tmp/segtest/0.jpg"); /* 0 */
box = boxCreate(105, 161, 620, 872); /* image region */
boxa1 = boxaCreate(1);
boxaAddBox(boxa1, box, L_INSERT);
boxaaAddBoxa(baa, boxa1, L_INSERT);
pixDestroy(&pix1);
pixDestroy(&pix2);
/* Compute image region at w = 2 * WIDTH */
pix1 = pixRead("candelabrum-11.jpg");
pix2 = pixScaleToSize(pix1, WIDTH, 0);
pix3 = pixConvertTo1(pix2, 100);
pix4 = pixExpandBinaryPower2(pix3, 2); /* w = 2 * WIDTH */
pix5 = pixGenHalftoneMask(pix4, NULL, NULL, 1);
pix6 = pixMorphSequence(pix5, "c20.1 + c1.20", 0);
pix7 = pixMaskConnComp(pix6, 8, &boxa1);
pix8 = pixReduceBinary2(pix7, NULL); /* back to w = WIDTH */
pix9 = pixBackgroundNormSimple(pix2, pix8, NULL);
pixWrite("/tmp/segtest/1.jpg", pix9, IFF_JFIF_JPEG);
regTestCheckFile(rp, "/tmp/segtest/1.jpg"); /* 1 */
boxa2 = boxaTransform(boxa1, 0, 0, 0.5, 0.5); /* back to w = WIDTH */
boxaaAddBoxa(baa, boxa2, L_INSERT);
pixDestroy(&pix1);
pixDestroy(&pix2);
pixDestroy(&pix3);
pixDestroy(&pix4);
pixDestroy(&pix5);
pixDestroy(&pix6);
pixDestroy(&pix7);
pixDestroy(&pix8);
pixDestroy(&pix9);
boxaDestroy(&boxa1);
/* Use mask to find image region */
pix1 = pixRead("lion-page.00016.jpg");
pix2 = pixScaleToSize(pix1, WIDTH, 0);
pixWrite("/tmp/segtest/2.jpg", pix2, IFF_JFIF_JPEG);
regTestCheckFile(rp, "/tmp/segtest/2.jpg"); /* 2 */
pix3 = pixRead("lion-mask.00016.tif");
pix4 = pixScaleToSize(pix3, WIDTH, 0);
boxa1 = pixConnComp(pix4, NULL, 8);
boxaaAddBoxa(baa, boxa1, L_INSERT);
pixDestroy(&pix1);
pixDestroy(&pix2);
pixDestroy(&pix3);
pixDestroy(&pix4);
/* Compute image region at full res */
pix1 = pixRead("rabi.png");
scalefactor = (l_float32)WIDTH / (l_float32)pixGetWidth(pix1);
pix2 = pixScaleToGray(pix1, scalefactor);
pixWrite("/tmp/segtest/3.jpg", pix2, IFF_JFIF_JPEG);
regTestCheckFile(rp, "/tmp/segtest/3.jpg"); /* 3 */
pix3 = pixGenHalftoneMask(pix1, NULL, NULL, 0);
pix4 = pixMorphSequence(pix3, "c20.1 + c1.20", 0);
boxa1 = pixConnComp(pix4, NULL, 8);
boxa2 = boxaTransform(boxa1, 0, 0, scalefactor, scalefactor);
boxaaAddBoxa(baa, boxa2, L_INSERT);
pixDestroy(&pix1);
pixDestroy(&pix2);
pixDestroy(&pix3);
pixDestroy(&pix4);
boxaDestroy(&boxa1);
/* Page with no image regions */
pix1 = pixRead("lucasta-47.jpg");
pix2 = pixScaleToSize(pix1, WIDTH, 0);
boxa1 = boxaCreate(1);
pixWrite("/tmp/segtest/4.jpg", pix2, IFF_JFIF_JPEG);
regTestCheckFile(rp, "/tmp/segtest/4.jpg"); /* 4 */
boxaaAddBoxa(baa, boxa1, L_INSERT);
pixDestroy(&pix1);
pixDestroy(&pix2);
/* Page that is all image */
pix1 = pixRead("map1.jpg");
pix2 = pixScaleToSize(pix1, WIDTH, 0);
pixWrite("/tmp/segtest/5.jpg", pix2, IFF_JFIF_JPEG);
regTestCheckFile(rp, "/tmp/segtest/5.jpg"); /* 5 */
h = pixGetHeight(pix2);
box = boxCreate(0, 0, WIDTH, h);
boxa1 = boxaCreate(1);
boxaAddBox(boxa1, box, L_INSERT);
boxaaAddBoxa(baa, boxa1, L_INSERT);
pixDestroy(&pix1);
pixDestroy(&pix2);
/* Save the boxaa file */
boxaaWrite("/tmp/segtest/seg.baa", baa);
regTestCheckFile(rp, "/tmp/segtest/seg.baa"); /* 6 */
/* Do the conversion */
l_pdfSetDateAndVersion(FALSE);
convertSegmentedFilesToPdf("/tmp/segtest", ".jpg", 100, L_G4_ENCODE,
140, baa, 75, 0.6, "Segmentation Test",
"/tmp/pdfseg.7.pdf");
regTestCheckFile(rp, "/tmp/pdfseg.7.pdf"); /* 7 */
boxaaDestroy(&baa);
return regTestCleanup(rp);
}
main(int argc,
char **argv)
{
char *pagedir, *pagesubstr, *maskdir, *masksubstr;
char *title, *fileout, *boxaafile, *boxaapath;
l_int32 ret, res, type, thresh;
l_float32 scalefactor;
BOXAA *baa;
static char mainName[] = "convertsegfilestopdf";
if (argc != 12) {
fprintf(stderr,
" Syntax: convertsegfilestopdf dirin substr res type thresh \\ \n"
" boxaafile scalefactor title fileout\n"
" where\n"
" pagedir: input directory for image files\n"
" pagesubstr: Use 'allfiles' to convert all files\n"
" in the directory\n"
" maskdir: input directory for mask files;\n"
" use 'skip' to skip \n"
" masksubstr: Use 'allfiles' to convert all files\n"
" in the directory; 'skip' to skip\n"
" res: Input resolution of each image;\n"
" assumed to all be the same\n"
" type: compression used for non-image regions:\n"
" 0: default (G4 encoding)\n"
" 1: JPEG encoding\n"
" 2: G4 encoding\n"
" 3: PNG encoding\n"
" thresh: threshold for binarization; use 0 for default\n"
" boxaafile: Optional file of 'image' regions within\n"
" each page. This contains a boxa for each\n"
" page, consisting of a set of regions.\n"
" Use 'skip' to skip.\n"
" scalefactor: Use to scale down the image regions\n"
" title: Use 'none' to omit\n"
" fileout: Output pdf file\n");
return 1;
}
pagedir = argv[1];
pagesubstr = argv[2];
maskdir = argv[3];
masksubstr = argv[4];
res = atoi(argv[5]);
type = atoi(argv[6]);
thresh = atoi(argv[7]);
boxaafile = argv[8];
scalefactor = atof(argv[9]);
title = argv[10];
fileout = argv[11];
if (!strcmp(pagesubstr, "allfiles"))
pagesubstr = NULL;
if (!strcmp(maskdir, "skip"))
maskdir = NULL;
if (!strcmp(masksubstr, "allfiles"))
masksubstr = NULL;
if (scalefactor <= 0.0 || scalefactor > 1.0) {
L_WARNING("invalid scalefactor: setting to 1.0", mainName);
scalefactor = 1.0;
}
if (type != 1 && type != 2 && type != 3)
type = L_G4_ENCODE;
if (thresh <= 0)
thresh = 150;
if (!strcmp(title, "none"))
title = NULL;
if (maskdir) /* use this; ignore any input boxaafile */
baa = convertNumberedMasksToBoxaa(maskdir, masksubstr, 0, 0);
else if (strcmp(boxaafile, "skip")) { /* use the boxaafile */
boxaapath = genPathname(boxaafile, NULL);
baa = boxaaRead(boxaapath);
FREE(boxaapath);
}
else /* no maskdir and no input boxaafile */
baa = NULL;
ret = convertSegmentedFilesToPdf(pagedir, pagesubstr, res, type, thresh,
baa, 75, scalefactor, title, fileout);
boxaaDestroy(&baa);
return ret;
}
