255 lines
9 KiB
Diff
255 lines
9 KiB
Diff
|
From 8c11a20925686855023df90ed477957c7d7fe91e Mon Sep 17 00:00:00 2001
|
||
|
|
From: Daniel Fullmer <danielrf12@gmail.com>
|
||
|
|
Date: Fri, 13 Sep 2019 15:54:21 -0400
|
||
|
|
Subject: [PATCH] Willus mod for k2pdfopt
|
||
|
|
|
||
|
|
---
|
||
|
|
src/allheaders.h | 4 ++
|
||
|
|
src/dewarp2.c | 106 ++++++++++++++++++++++++++++++++++++++++++-----
|
||
|
|
src/leptwin.c | 6 ++-
|
||
|
|
3 files changed, 104 insertions(+), 12 deletions(-)
|
||
|
|
|
||
|
|
diff --git a/src/allheaders.h b/src/allheaders.h
|
||
|
|
index e68eff1..b3cc729 100644
|
||
|
|
--- a/src/allheaders.h
|
||
|
|
+++ b/src/allheaders.h
|
||
|
|
@@ -669,6 +669,10 @@ LEPT_DLL extern L_DEWARPA * dewarpaReadMem ( const l_uint8 *data, size_t size );
|
||
|
|
LEPT_DLL extern l_ok dewarpaWrite ( const char *filename, L_DEWARPA *dewa );
|
||
|
|
LEPT_DLL extern l_ok dewarpaWriteStream ( FILE *fp, L_DEWARPA *dewa );
|
||
|
|
LEPT_DLL extern l_ok dewarpaWriteMem ( l_uint8 **pdata, size_t *psize, L_DEWARPA *dewa );
|
||
|
|
+/* WILLUS MOD */
|
||
|
|
+ LEPT_DLL extern l_int32 dewarpBuildPageModel_ex ( L_DEWARP *dew, const char *debugfile,l_int32 fit_order );
|
||
|
|
+ LEPT_DLL extern l_int32 dewarpFindVertDisparity_ex ( L_DEWARP *dew, PTAA *ptaa, l_int32 rotflag,l_int32 fit_order );
|
||
|
|
+ LEPT_DLL extern l_int32 dewarpBuildLineModel_ex ( L_DEWARP *dew, l_int32 opensize, const char *debugfile,l_int32 fit_order );
|
||
|
|
LEPT_DLL extern l_ok dewarpBuildPageModel ( L_DEWARP *dew, const char *debugfile );
|
||
|
|
LEPT_DLL extern l_ok dewarpFindVertDisparity ( L_DEWARP *dew, PTAA *ptaa, l_int32 rotflag );
|
||
|
|
LEPT_DLL extern l_ok dewarpFindHorizDisparity ( L_DEWARP *dew, PTAA *ptaa );
|
||
|
|
diff --git a/src/dewarp2.c b/src/dewarp2.c
|
||
|
|
index 220eec1..2e29500 100644
|
||
|
|
--- a/src/dewarp2.c
|
||
|
|
+++ b/src/dewarp2.c
|
||
|
|
@@ -144,9 +144,17 @@ static const l_float32 L_ALLOWED_W_FRACT = 0.05; /* no bigger */
|
||
|
|
* longest textlines.
|
||
|
|
* </pre>
|
||
|
|
*/
|
||
|
|
+/* WILLUS MOD */
|
||
|
|
l_ok
|
||
|
|
-dewarpBuildPageModel(L_DEWARP *dew,
|
||
|
|
- const char *debugfile)
|
||
|
|
+dewarpBuildPageModel(L_DEWARP *dew,const char *debugfile)
|
||
|
|
+{
|
||
|
|
+return(dewarpBuildPageModel_ex(dew,debugfile,2));
|
||
|
|
+}
|
||
|
|
+
|
||
|
|
+l_ok
|
||
|
|
+dewarpBuildPageModel_ex(L_DEWARP *dew,
|
||
|
|
+ const char *debugfile,
|
||
|
|
+ l_int32 fit_order)
|
||
|
|
{
|
||
|
|
l_int32 linecount, topline, botline, ret;
|
||
|
|
PIX *pixs, *pix1, *pix2, *pix3;
|
||
|
|
@@ -225,7 +233,7 @@ PTAA *ptaa1, *ptaa2;
|
||
|
|
/* Get the sampled vertical disparity from the textline centers.
|
||
|
|
* The disparity array will push pixels vertically so that each
|
||
|
|
* textline is flat and centered at the y-position of the mid-point. */
|
||
|
|
- if (dewarpFindVertDisparity(dew, ptaa2, 0) != 0) {
|
||
|
|
+ if (dewarpFindVertDisparity_ex(dew, ptaa2, 0, fit_order) != 0) {
|
||
|
|
L_WARNING("vertical disparity not built\n", procName);
|
||
|
|
ptaaDestroy(&ptaa2);
|
||
|
|
return 1;
|
||
|
|
@@ -290,13 +298,24 @@ PTAA *ptaa1, *ptaa2;
|
||
|
|
* a pdf. Non-pix debug output goes to /tmp.
|
||
|
|
* </pre>
|
||
|
|
*/
|
||
|
|
+/* WILLUS MOD */
|
||
|
|
l_ok
|
||
|
|
dewarpFindVertDisparity(L_DEWARP *dew,
|
||
|
|
PTAA *ptaa,
|
||
|
|
l_int32 rotflag)
|
||
|
|
{
|
||
|
|
+return(dewarpFindVertDisparity_ex(dew,ptaa,rotflag,2));
|
||
|
|
+}
|
||
|
|
+/* WILLUS MOD -- add cubic and quartic fits and ..._ex functions */
|
||
|
|
+l_int32
|
||
|
|
+dewarpFindVertDisparity_ex(L_DEWARP *dew,
|
||
|
|
+ PTAA *ptaa,
|
||
|
|
+ l_int32 rotflag,
|
||
|
|
+ l_int32 fit_order)
|
||
|
|
+{
|
||
|
|
l_int32 i, j, nlines, npts, nx, ny, sampling;
|
||
|
|
-l_float32 c0, c1, c2, x, y, midy, val, medval, meddev, minval, maxval;
|
||
|
|
+/* WILLUS MOD */
|
||
|
|
+l_float32 c0, c1, c2, c3, c4, x, y, midy, val, medval, meddev, minval, maxval;
|
||
|
|
l_float32 *famidys;
|
||
|
|
NUMA *nax, *nafit, *nacurve0, *nacurve1, *nacurves;
|
||
|
|
NUMA *namidy, *namidys, *namidysi;
|
||
|
|
@@ -304,11 +323,22 @@ PIX *pix1, *pix2, *pixcirc, *pixdb;
|
||
|
|
PTA *pta, *ptad, *ptacirc;
|
||
|
|
PTAA *ptaa0, *ptaa1, *ptaa2, *ptaa3, *ptaa4, *ptaa5, *ptaat;
|
||
|
|
FPIX *fpix;
|
||
|
|
+/* WILLUS MOD */
|
||
|
|
+l_int32 fit_order1,fit_order2;
|
||
|
|
|
||
|
|
PROCNAME("dewarpFindVertDisparity");
|
||
|
|
|
||
|
|
if (!dew)
|
||
|
|
return ERROR_INT("dew not defined", procName, 1);
|
||
|
|
+/* WILLUS MOD */
|
||
|
|
+ if (fit_order < 10)
|
||
|
|
+ fit_order1 = fit_order2 = fit_order;
|
||
|
|
+ else
|
||
|
|
+ {
|
||
|
|
+ fit_order1=fit_order % 10;
|
||
|
|
+ fit_order2=fit_order / 10;
|
||
|
|
+ fit_order2=fit_order2 % 10;
|
||
|
|
+ }
|
||
|
|
dew->vsuccess = 0;
|
||
|
|
if (!ptaa)
|
||
|
|
return ERROR_INT("ptaa not defined", procName, 1);
|
||
|
|
@@ -331,12 +361,32 @@ FPIX *fpix;
|
||
|
|
pixdb = (rotflag) ? pixRotateOrth(dew->pixs, 1) : pixClone(dew->pixs);
|
||
|
|
for (i = 0; i < nlines; i++) { /* for each line */
|
||
|
|
pta = ptaaGetPta(ptaa, i, L_CLONE);
|
||
|
|
- ptaGetQuadraticLSF(pta, &c2, &c1, &c0, NULL);
|
||
|
|
- numaAddNumber(nacurve0, c2);
|
||
|
|
+/* WILLUS MOD */
|
||
|
|
+if (fit_order1>3)
|
||
|
|
+ {
|
||
|
|
+ ptaGetQuarticLSF(pta, &c4, &c3, &c2, &c1, &c0, NULL);
|
||
|
|
+ numaAddNumber(nacurve0, c4);
|
||
|
|
+ }
|
||
|
|
+else if (fit_order1==3)
|
||
|
|
+ {
|
||
|
|
+ ptaGetCubicLSF(pta, &c3, &c2, &c1, &c0, NULL);
|
||
|
|
+ numaAddNumber(nacurve0, c3);
|
||
|
|
+ }
|
||
|
|
+else
|
||
|
|
+ {
|
||
|
|
+ ptaGetQuadraticLSF(pta, &c2, &c1, &c0, NULL);
|
||
|
|
+ numaAddNumber(nacurve0, c2);
|
||
|
|
+ }
|
||
|
|
ptad = ptaCreate(nx);
|
||
|
|
for (j = 0; j < nx; j++) { /* uniformly sampled in x */
|
||
|
|
x = j * sampling;
|
||
|
|
- applyQuadraticFit(c2, c1, c0, x, &y);
|
||
|
|
+/* WILLUS MOD */
|
||
|
|
+if (fit_order1>3)
|
||
|
|
+ applyQuarticFit(c4, c3, c2, c1, c0, x, &y);
|
||
|
|
+else if (fit_order1==3)
|
||
|
|
+ applyCubicFit(c3, c2, c1, c0, x, &y);
|
||
|
|
+else
|
||
|
|
+ applyQuadraticFit(c2, c1, c0, x, &y);
|
||
|
|
ptaAddPt(ptad, x, y);
|
||
|
|
}
|
||
|
|
ptaaAddPta(ptaa0, ptad, L_INSERT);
|
||
|
|
@@ -350,7 +400,13 @@ FPIX *fpix;
|
||
|
|
for (i = 0; i < nlines; i++) {
|
||
|
|
pta = ptaaGetPta(ptaa, i, L_CLONE);
|
||
|
|
ptaGetArrays(pta, &nax, NULL);
|
||
|
|
- ptaGetQuadraticLSF(pta, NULL, NULL, NULL, &nafit);
|
||
|
|
+/* WILLUS MOD */
|
||
|
|
+if (fit_order1>3)
|
||
|
|
+ptaGetQuarticLSF(pta, NULL, NULL, NULL, NULL, NULL, &nafit);
|
||
|
|
+else if (fit_order1==3)
|
||
|
|
+ptaGetCubicLSF(pta, NULL, NULL, NULL, NULL, &nafit);
|
||
|
|
+else
|
||
|
|
+ptaGetQuadraticLSF(pta, NULL, NULL, NULL, &nafit);
|
||
|
|
ptad = ptaCreateFromNuma(nax, nafit);
|
||
|
|
ptaaAddPta(ptaat, ptad, L_INSERT);
|
||
|
|
ptaDestroy(&pta);
|
||
|
|
@@ -494,11 +550,24 @@ FPIX *fpix;
|
||
|
|
ptaa5 = ptaaCreate(nx); /* uniformly sampled across full height of image */
|
||
|
|
for (j = 0; j < nx; j++) { /* for each column */
|
||
|
|
pta = ptaaGetPta(ptaa4, j, L_CLONE);
|
||
|
|
- ptaGetQuadraticLSF(pta, &c2, &c1, &c0, NULL);
|
||
|
|
+/* WILLUS MOD */
|
||
|
|
+/* Order higher than 2 can cause a little craziness here. */
|
||
|
|
+if (fit_order2>3)
|
||
|
|
+ ptaGetQuarticLSF(pta, &c4, &c3, &c2, &c1, &c0, NULL);
|
||
|
|
+else if (fit_order2==3)
|
||
|
|
+ ptaGetCubicLSF(pta, &c3, &c2, &c1, &c0, NULL);
|
||
|
|
+else
|
||
|
|
+ ptaGetQuadraticLSF(pta, &c2, &c1, &c0, NULL);
|
||
|
|
ptad = ptaCreate(ny);
|
||
|
|
for (i = 0; i < ny; i++) { /* uniformly sampled in y */
|
||
|
|
y = i * sampling;
|
||
|
|
- applyQuadraticFit(c2, c1, c0, y, &val);
|
||
|
|
+/* WILLUS MOD */
|
||
|
|
+if (fit_order2>3)
|
||
|
|
+ applyQuarticFit(c4, c3, c2, c1, c0, y, &val);
|
||
|
|
+else if (fit_order2==3)
|
||
|
|
+ applyCubicFit(c3, c2, c1, c0, y, &val);
|
||
|
|
+else
|
||
|
|
+ applyQuadraticFit(c2, c1, c0, y, &val);
|
||
|
|
ptaAddPt(ptad, y, val);
|
||
|
|
}
|
||
|
|
ptaaAddPta(ptaa5, ptad, L_INSERT);
|
||
|
|
@@ -1602,11 +1671,21 @@ FPIX *fpix;
|
||
|
|
* See notes there.
|
||
|
|
* </pre>
|
||
|
|
*/
|
||
|
|
+/* WILLUS MOD */
|
||
|
|
l_ok
|
||
|
|
dewarpBuildLineModel(L_DEWARP *dew,
|
||
|
|
l_int32 opensize,
|
||
|
|
const char *debugfile)
|
||
|
|
{
|
||
|
|
+return(dewarpBuildLineModel_ex(dew,opensize,debugfile,2));
|
||
|
|
+}
|
||
|
|
+
|
||
|
|
+l_int32
|
||
|
|
+dewarpBuildLineModel_ex(L_DEWARP *dew,
|
||
|
|
+ l_int32 opensize,
|
||
|
|
+ const char *debugfile,
|
||
|
|
+ l_int32 fit_order)
|
||
|
|
+{
|
||
|
|
char buf[64];
|
||
|
|
l_int32 i, j, bx, by, ret, nlines;
|
||
|
|
BOXA *boxa;
|
||
|
|
@@ -1695,6 +1774,8 @@ PTAA *ptaa1, *ptaa2;
|
||
|
|
|
||
|
|
/* Remove all lines that are not at least 0.75 times the length
|
||
|
|
* of the longest line. */
|
||
|
|
+/* WILLUS MOD */
|
||
|
|
+/*
|
||
|
|
ptaa2 = dewarpRemoveShortLines(pix, ptaa1, 0.75, DEBUG_SHORT_LINES);
|
||
|
|
if (debugfile) {
|
||
|
|
pix1 = pixConvertTo32(pix);
|
||
|
|
@@ -1704,6 +1785,8 @@ PTAA *ptaa1, *ptaa2;
|
||
|
|
pixDestroy(&pix1);
|
||
|
|
pixDestroy(&pix2);
|
||
|
|
}
|
||
|
|
+*/
|
||
|
|
+ptaa2=ptaa1;
|
||
|
|
ptaaDestroy(&ptaa1);
|
||
|
|
nlines = ptaaGetCount(ptaa2);
|
||
|
|
if (nlines < dew->minlines) {
|
||
|
|
@@ -1717,7 +1800,8 @@ PTAA *ptaa1, *ptaa2;
|
||
|
|
* centers. The disparity array will push pixels vertically
|
||
|
|
* so that each line is flat and centered at the y-position
|
||
|
|
* of the mid-point. */
|
||
|
|
- ret = dewarpFindVertDisparity(dew, ptaa2, 1 - i);
|
||
|
|
+/* WILLUS MOD */
|
||
|
|
+ ret = dewarpFindVertDisparity_ex(dew, ptaa2, 1 - i, fit_order);
|
||
|
|
|
||
|
|
/* If i == 0, move the result to the horizontal disparity,
|
||
|
|
* rotating it back by -90 degrees. */
|
||
|
|
diff --git a/src/leptwin.c b/src/leptwin.c
|
||
|
|
index 72643a0..573d33e 100644
|
||
|
|
--- a/src/leptwin.c
|
||
|
|
+++ b/src/leptwin.c
|
||
|
|
@@ -364,5 +364,9 @@ PIXCMAP *cmap;
|
||
|
|
|
||
|
|
return hBitmap;
|
||
|
|
}
|
||
|
|
-
|
||
|
|
+#else
|
||
|
|
+/* willus mod: Avoid weird issue with OS/X library archiver when there are no symbols */
|
||
|
|
+int leptwin_my_empty_func(void);
|
||
|
|
+int leptwin_my_empty_func(void)
|
||
|
|
+{return(0);}
|
||
|
|
#endif /* _WIN32 */
|
||
|
|
--
|
||
|
|
2.22.0
|
||
|
|
|