antialiasing text conversion: canonimage has 0..7

diff --git a/pctb/convert.h b/pctb/convert.h
index b61563fd32449c200b6915484b4fcc9a3d3c7a33..0ab09dbccb7ac56b2076cb870d9956f701af6f09 100644 (file)
--- a/pctb/convert.h
+++ b/pctb/convert.h
@@ -69,7 +69,7 @@ static inline Rgb ri_rgb(const RgbImage *ri, int x, int y) {
 
 /*----- from structure.c -----*/
 
-void find_structure(const CanonImage *im, int *max_relevant_y_r);
+void find_structure(CanonImage *im, int *max_relevant_y_r);
 Rect find_sunshine_widget(void);
 
 void find_islandname(RgbImage *ri);

diff --git a/pctb/structure.c b/pctb/structure.c
index 1f3c743319de029d342bbdbaa9bce1ec920047cb..a3cca0b4408fc3da115bd91f069589956c0fb0f7 100644 (file)
--- a/pctb/structure.c
+++ b/pctb/structure.c
@@ -27,7 +27,7 @@
 
 #include "structure.h"
 
-static const CanonImage *cim;
+static CanonImage *cim;
 
 static inline char get(int x, int y) { return cim->d[y * cim->w + x]; }
 static inline char get_p(Point p) { return get(p.x,p.y); }
@@ -123,8 +123,11 @@ static void mustfail2(void) {
 
 #define MP(v) fprintf(stderr," %s=%d,%d",#v,(v).x,(v).y)
 #define MI(v) fprintf(stderr," %s=%d",   #v,(v))
+#define MIL(v) fprintf(stderr," %s=%ld", #v,(v))
+#define MRGB(v) fprintf(stderr," %s=%06lx", #v,(v))
 #define MC(v) fprintf(stderr," %s='%c'", #v,(v))
 #define MS(v) fprintf(stderr," %s=\"%s\"", #v,(v))
+#define MF(v) fprintf(stderr," %s=\"%f\"", #v,(v))
 #define MSB(v) fprintf(stderr," %s", (v))
 #define MR(v) fprintf(stderr," %s=%d,%d..%d,%d",\
                       #v,(v).tl.x,(v).tl.y,(v).br.x,(v).br.y)
@@ -133,7 +136,7 @@ static void mustfail2(void) {
 #define REQUIRE_RECTANGLE(tlx,tly,brx,bry,ok) \
  require_rectangle(tlx, tly, brx, bry, ok, __LINE__);
 
-#define FOR_P_RECT(p,rect)                             \
+#define FOR_P_RECT(p,rr)                               \
   for ((p).x=(rr).tl.x; (p).x<=(rr).br.x; (p).x++)     \
     for ((p).y=(rr).tl.y; (p).y<=(rr).br.y; (p).y++)
 
@@ -224,7 +227,7 @@ static int commod_selector_matches(Rect search, const char *const *all,
     (search).TLBR.XY += increm;                                                   \
   }
 
-void find_structure(const CanonImage *im, int *max_relevant_y_r) {
+void find_structure(CanonImage *im, int *max_relevant_y_r) {
   cim= im;
   
   Rect whole = { {0,0}, {cim->w-1,cim->h-1} };
@@ -447,13 +450,73 @@ static void find_commodity(int offset, Rect *rr) {
     REQUIRE_RECTANGLE(rr->tl.x,rr->br.y+1, rr->br.x,rr->br.y+1, "+");
 }
 
-static void find_table_entry(Rect commod, int colno, Rect *cellr) {
-  cellr->tl.y= commod.tl.y;
-  cellr->br.y= commod.br.y;
-  cellr->tl.x= !colno ? commod.tl.x : colrightx[colno-1]+2;
-  cellr->br.x=                        colrightx[colno];
-  debug_rect("cell", colno, *cellr);
-  require_rectangle_r(*cellr, " o", __LINE__);
+static void find_table_entry(Rect commod, int colno, Rect *cell) {
+  cell->tl.y= commod.tl.y;
+  cell->br.y= commod.br.y;
+  cell->tl.x= !colno ? commod.tl.x : colrightx[colno-1]+2;
+  cell->br.x=                        colrightx[colno];
+  debug_rect("cell", colno, *cell);
+
+  const RgbImage *ri= cim->rgb;
+  
+  Rgb background= ri_rgb(ri, cell->br.x, cell->br.y);
+  long bg_count=0, white_count=0, black_count=0;
+  Point p;
+  FOR_P_RECT(p,*cell) {
+    Rgb here= ri_rgb(ri, p.x, p.y);
+    if (here == background) bg_count++;
+    else if (here == 0) black_count++;
+    else if (here == 0xffffffU) white_count++;
+  }
+  long total_count= RECT_W(*cell) * RECT_H(*cell);
+  MUST( bg_count > 8*total_count / 10,
+       MR(*cell);MIL(bg_count);MIL(white_count);MIL(black_count) );
+  if (bg_count == total_count)
+    return;
+
+  MUST( !!black_count != !!white_count,
+       MR(*cell);MIL(bg_count);MIL(white_count);MIL(black_count) );
+
+  debugf("TABLEENTRY col=%d %d,%d..%d,%d bg=%ld white=%ld black=%ld\n",
+        colno, cell->tl.x,cell->tl.y, cell->br.x,cell->br.y,
+        bg_count, white_count, black_count);
+  
+  Rgb foreground= white_count ? 0xffffffU : 0;
+  int monochrome= 1;
+
+  FOR_P_RECT(p,*cell) {
+    Rgb here= ri_rgb(ri, p.x, p.y);
+    double alpha[3], alpha_mean=0;
+    int i;
+    for (i=0; i<3; i++) {
+      unsigned char here_chan= here       >> (i*8);
+      unsigned char bg_chan=   background >> (i*8);
+      unsigned char fg_chan=   foreground >> (i*8);
+      double alpha_chan=
+       ((double)here_chan - (double)bg_chan) /
+       ((double)fg_chan   - (double)bg_chan);
+      alpha[i]= alpha_chan;
+      alpha_mean += alpha_chan / 3;
+    }
+
+    double alpha_min = alpha_mean - 0.13;
+    double alpha_max = alpha_mean + 0.13;
+    for (i=0; i<3; i++)
+      MUST( alpha_min <= alpha[i] && alpha[i] <= alpha_max,
+           MI(i);MRGB(here);MRGB(background);MRGB(foreground);
+           MF(alpha_min);MF(alpha[i]);MF(alpha_max) );
+
+    MUST( 0 <= alpha_mean && alpha_mean <= 1,
+         MRGB(here);MRGB(background);MRGB(foreground);
+         MF(alpha_mean) );
+    int here_int= 7*alpha_mean;
+    if (!(here_int==0 || here_int==7)) monochrome=0;
+    cim->d[p.y * cim->w + p.x]= '0' + here_int;
+  }
+
+  debug_rect("cell07", colno, *cell);
+
+  require_rectangle_r(*cell, "0123456789", __LINE__);
 }
 
 static void ocr_rectangle(Rect r, const OcrCellType ct, FILE *tsv_output) {
@@ -469,7 +532,8 @@ static void ocr_rectangle(Rect r, const OcrCellType ct, FILE *tsv_output) {
       int pixel= get_p(here);
       switch (pixel) {
       case ' ':           break;
-      case 'o': cx |= rv; break;
+      case '0':           break;
+      case '7': cx |= rv; break;
       default:
        MUST(!"wrong pixel",
             MC(pixel);MP(here);MSB(ocr_celltype_name(ct));MR(r); );
@@ -692,5 +756,4 @@ void find_islandname(RgbImage *ri) {
   archipelago= masprintf("%.*s", (int)(delim-archisland), archisland);
   island= masprintf("%s", delim+3);
 
-  free(ri);
 }