ho_bitmap *
ho_bitmap_filter_fill (const ho_bitmap * m)
{
ho_objmap *m_obj;
ho_bitmap *m_out;
ho_bitmap *m_temp1;
ho_bitmap *m_temp2;
int index;
int width, height;
/* allocate memory */
m_obj = ho_objmap_new_from_bitmap (m);
if (!m_obj)
return NULL;
m_out = ho_bitmap_new (m->width, m->height);
if (!m_out)
{
ho_objmap_free (m_obj);
return NULL;
}
m_out->x = m->x;
m_out->y = m->y;
m_out->type = m->type;
m_out->font_height = m->font_height;
m_out->font_width = m->font_width;
m_out->font_spacing = m->font_spacing;
m_out->line_spacing = m->line_spacing;
m_out->avg_line_fill = m->avg_line_fill;
m_out->com_line_fill = m->com_line_fill;
m_out->nikud = m->nikud;
/* loop over all the objects and box them */
for (index = 0; index < m_obj->obj_list->size; index++)
{
/* get a dimention factor */
width = (((m_obj->obj_list)->objects)[index]).width;
height = (((m_obj->obj_list)->objects)[index]).height;
/* copy only the current object to a new bitmap */
m_temp1 = ho_objmap_to_bitmap_by_index (m_obj, index);
/* fill the current object */
m_temp2 = ho_bitmap_hlink (m_temp1, width / 4);
ho_bitmap_free (m_temp1);
m_temp1 = ho_bitmap_vlink (m_temp2, height / 4);
ho_bitmap_free (m_temp2);
/* add to matrix out */
ho_bitmap_or (m_out, m_temp1);
ho_bitmap_free (m_temp1);
}
return m_out;
}
ho_bitmap *
ho_bitmap_filter_set_height_from_bottom (const ho_bitmap * m,
const int height, const int top, const int bottom)
{
ho_objmap *m_obj;
ho_bitmap *m_out;
ho_bitmap *m_temp1;
ho_bitmap *m_temp2;
int index;
/* allocate memory */
m_obj = ho_objmap_new_from_bitmap (m);
if (!m_obj)
return NULL;
m_out = ho_bitmap_new (m->width, m->height);
if (!m_out)
{
ho_objmap_free (m_obj);
return NULL;
}
m_out->x = m->x;
m_out->y = m->y;
m_out->type = m->type;
m_out->font_height = m->font_height;
m_out->font_width = m->font_width;
m_out->font_spacing = m->font_spacing;
m_out->line_spacing = m->line_spacing;
m_out->avg_line_fill = m->avg_line_fill;
m_out->com_line_fill = m->com_line_fill;
m_out->nikud = m->nikud;
/* loop over all the objects and box them */
for (index = 0; index < m_obj->obj_list->size; index++)
{
/* copy only the current object to a new bitmap */
m_temp1 = ho_objmap_to_bitmap_by_index (m_obj, index);
if (!m_temp1)
continue;
/* take height pixels from this object */
m_temp2 = ho_bitmap_set_height_from_bottom (m_temp1, height, top, bottom);
ho_bitmap_free (m_temp1);
if (!m_temp2)
continue;
/* add to matrix out */
ho_bitmap_or (m_out, m_temp2);
ho_bitmap_free (m_temp2);
}
return m_out;
}
ho_bitmap *
ho_bitmap_filter_boxes (const ho_bitmap * m, const int leeway_down,
const int leeway_up)
{
ho_objmap *m_obj;
ho_bitmap *m_out;
int index;
int x, y, width, height;
/* allocate memory */
m_obj = ho_objmap_new_from_bitmap (m);
if (!m_obj)
return NULL;
m_out = ho_bitmap_new (m->width, m->height);
m_out->x = m->x;
m_out->y = m->y;
m_out->type = m->type;
m_out->font_height = m->font_height;
m_out->font_width = m->font_width;
m_out->font_spacing = m->font_spacing;
m_out->line_spacing = m->line_spacing;
m_out->avg_line_fill = m->avg_line_fill;
m_out->com_line_fill = m->com_line_fill;
m_out->nikud = m->nikud;
if (!m_out)
{
ho_objmap_free (m_obj);
return NULL;
}
/* loop over all the objects and box them */
for (index = 0; index < m_obj->obj_list->size; index++)
{
x = (((m_obj->obj_list)->objects)[index]).x;
y = (((m_obj->obj_list)->objects)[index]).y;
width = (((m_obj->obj_list)->objects)[index]).width;
height = (((m_obj->obj_list)->objects)[index]).height;
y -= leeway_up;
height += leeway_up + leeway_down;
if (y < 0)
y = 0;
if (y + height >= m->height)
height = m->height - y - 1;
ho_bitmap_draw_box (m_out, x, y, width, height);
}
return m_out;
}
int
ho_bitmap_filter_count_objects (const ho_bitmap * m)
{
ho_objmap *o_obj;
int count;
o_obj = ho_objmap_new_from_bitmap (m);
if (!o_obj)
return -1;
count = ho_objmap_get_size (o_obj);
ho_objmap_free (o_obj);
return count;
}
ho_bitmap *
ho_bitmap_filter_by_size (const ho_bitmap * m,
int min_height, int max_height, int min_width, int max_width)
{
ho_objmap *m_obj;
ho_bitmap *m_out;
/* create a new objmap */
m_obj = ho_objmap_new_from_bitmap (m);
if (!m_obj)
return NULL;
m_out = ho_objmap_to_bitmap_by_size (m_obj,
min_height, max_height, min_width, max_width);
/* free objmap */
ho_objmap_free (m_obj);
return m_out;
}
ho_bitmap *
ho_bitmap_filter_obj_extend_lateraly (const ho_bitmap * m, const int ext_width)
{
ho_objmap *m_obj;
ho_bitmap *m_temp;
ho_bitmap *m_out;
int x, y;
int index;
int width, height;
m_temp = ho_bitmap_clone (m);
if (!m_temp)
return NULL;
/* loop over all objects and extend them lateraly */
/* allocate memory */
m_obj = ho_objmap_new_from_bitmap (m_temp);
if (!m_obj)
{
ho_bitmap_free (m_temp);
return NULL;
}
/* draw stopers */
for (index = 0; index < m_obj->obj_list->size; index++)
{
x = (((m_obj->obj_list)->objects)[index]).x;
y = (((m_obj->obj_list)->objects)[index]).y;
width = (((m_obj->obj_list)->objects)[index]).width;
height = (((m_obj->obj_list)->objects)[index]).height;
if (x - ext_width < 0)
x = ext_width;
if (x + width + ext_width >= m->width)
width = m->width - x - ext_width - 1;
ho_bitmap_draw_vline (m_temp, x - ext_width, y, height);
ho_bitmap_draw_vline (m_temp, x + width + ext_width, y, height);
}
/* extend */
m_out = ho_bitmap_hlink (m_temp, 7 * ext_width / 4);
ho_bitmap_free (m_temp);
if (!m_out)
return NULL;
/* delete stopers */
for (index = 0; index < m_obj->obj_list->size; index++)
{
x = (((m_obj->obj_list)->objects)[index]).x;
y = (((m_obj->obj_list)->objects)[index]).y;
width = (((m_obj->obj_list)->objects)[index]).width;
height = (((m_obj->obj_list)->objects)[index]).height;
if (x - ext_width < 0)
x = ext_width;
if (x + width + ext_width >= m->width)
width = m->width - x - ext_width - 1;
ho_bitmap_delete_vline (m_out, x - ext_width, y, height);
ho_bitmap_delete_vline (m_out, x + width + ext_width, y, height);
}
/* set origin */
m_out->x = m->x;
m_out->y = m->y;
m_out->type = m->type;
m_out->font_height = m->font_height;
m_out->font_width = m->font_width;
m_out->font_spacing = m->font_spacing;
m_out->line_spacing = m->line_spacing;
m_out->avg_line_fill = m->avg_line_fill;
m_out->com_line_fill = m->com_line_fill;
m_out->nikud = m->nikud;
ho_objmap_free (m_obj);
return m_out;
}
ho_bitmap *
ho_bitmap_filter_remove_dots (const ho_bitmap * m,
const unsigned char erosion_n, const unsigned char dilation_n)
{
int x, y;
unsigned char sum;
ho_bitmap *m_temp;
ho_bitmap *m_out;
ho_objmap *m_obj;
int index;
int width, height;
/* allocate memory */
m_out = ho_bitmap_new (m->width, m->height);
if (!m_out)
return NULL;
m_out->x = m->x;
m_out->y = m->y;
m_out->type = m->type;
m_out->font_height = m->font_height;
m_out->font_width = m->font_width;
m_out->font_spacing = m->font_spacing;
m_out->line_spacing = m->line_spacing;
m_out->avg_line_fill = m->avg_line_fill;
m_out->com_line_fill = m->com_line_fill;
m_out->nikud = m->nikud;
/* connect all the small dots */
m_temp = ho_bitmap_dilation (m);
/* check the size of objects */
m_obj = ho_objmap_new_from_bitmap (m_temp);
ho_bitmap_free (m_temp);
for (x = 1; x < (m->width - 1); x++)
for (y = 1; y < (m->height - 1); y++)
{
/* check the size of this pixel's object */
index = ho_objmap_get (m_obj, x, y);
if (index)
{
width = (((m_obj->obj_list)->objects)[index - 1]).width;
height = (((m_obj->obj_list)->objects)[index - 1]).height;
}
else
{
width = 0;
height = 0;
}
/* in a big object do erosion */
if (width > m->width / 4 || height > m->height / 4)
{
if (ho_bitmap_get (m, x, y)) /* black pixel */
{
sum = ho_bitmap_get (m, x - 1, y - 1) +
ho_bitmap_get (m, x - 1, y) +
ho_bitmap_get (m, x - 1, y + 1) +
ho_bitmap_get (m, x, y - 1) +
ho_bitmap_get (m, x, y + 1) +
ho_bitmap_get (m, x + 1, y - 1) +
ho_bitmap_get (m, x + 1, y) + ho_bitmap_get (m, x + 1, y + 1);
/* n number of white pixels or more */
if ((8 - sum) < erosion_n)
ho_bitmap_set (m_out, x, y);
}
}
else /* if in a small object do dilation */
{
if (!ho_bitmap_get (m, x, y)) /* white pixel */
{
sum = ho_bitmap_get (m, x - 1, y - 1) +
ho_bitmap_get (m, x - 1, y) +
ho_bitmap_get (m, x - 1, y + 1) +
ho_bitmap_get (m, x, y - 1) +
ho_bitmap_get (m, x, y + 1) +
ho_bitmap_get (m, x + 1, y - 1) +
ho_bitmap_get (m, x + 1, y) + ho_bitmap_get (m, x + 1, y + 1);
/* n number of black neighbors or more */
if (sum >= dilation_n)
ho_bitmap_set (m_out, x, y);
}
else
ho_bitmap_set (m_out, x, y);
}
}
ho_objmap_free (m_obj);
return m_out;
}
int
ho_recognize_nikud_dimentions (const ho_bitmap * m_text,
const ho_bitmap * m_mask, double *height,
double *width, double *top, double *bottom,
double *top_left, double *top_mid, double *top_right,
double *mid_left, double *mid_right,
double *bottom_left, double *bottom_mid,
double *bottom_right,
double *dots_above, double *dots_below, double *dots_inside,
double *objs_above, double *objs_below, double *objs_inside,
double *main_obj_height, double *main_obj_width,
double *top_dot_x, double *inside_dot_x, double *font_width_by_height)
{
int i, x, y;
int obj_width, obj_height;
int obj_x, obj_y;
int obj_w, obj_h;
int line_start, line_end, line_height;
int sum, font_start_x, font_end_x, font_height, font_width;
int font_start_y, font_end_y;
ho_objmap *o_obj = NULL;
unsigned char is_dot, is_obj;
/* init values to zero */
*height = 0.0;
*width = 0.0;
*top = 0.0;
*bottom = 0.0;
*top_left = 0.0;
*top_mid = 0.0;
*top_right = 0.0;
*mid_left = 0.0;
*mid_right = 0.0;
*bottom_left = 0.0;
*bottom_mid = 0.0;
*bottom_right = 0.0;
*dots_above = 0.0;
*dots_below = 0.0;
*dots_inside = 0.0;
*objs_above = 0.0;
*objs_below = 0.0;
*objs_inside = 0.0;
*top_dot_x = 0.0;
*inside_dot_x = 0.0;
*font_width_by_height = 0.0;
/* get line start and end */
x = m_mask->width / 2;
for (y = 0; y < m_mask->height && !ho_bitmap_get (m_mask, x, y); y++) ;
line_start = y - 1;
for (; y < m_mask->height && ho_bitmap_get (m_mask, x, y); y++) ;
line_end = y;
line_height = line_end - line_start;
if (line_height < 4 || m_text->width < 4)
return TRUE;
*font_width_by_height = (double) m_mask->width / (double) line_height;
/* get all the objects of nikud */
o_obj = ho_objmap_new_from_bitmap (m_text);
if (!o_obj)
return TRUE;
/* if no nikud just return */
if (ho_objmap_get_size (o_obj) == 0)
{
ho_objmap_free (o_obj);
return FALSE;
}
/* count points and objects above/in/below font */
obj_x = 0;
obj_y = 0;
obj_w = 0;
obj_h = 0;
for (i = 0; i < ho_objmap_get_size (o_obj); i++)
{
/* is this object inside line ? */
y = ho_objmap_get_object (o_obj, i).y;
x = ho_objmap_get_object (o_obj, i).x;
obj_width = ho_objmap_get_object (o_obj, i).width;
obj_height = ho_objmap_get_object (o_obj, i).height;
is_dot = (obj_height < line_height / 4 && obj_height > line_height / 16
&& obj_width < line_height / 4 && obj_width > line_height / 16);
is_obj = (obj_height > line_height / 4 || obj_width > line_height / 4);
if (is_dot)
{
/* count */
if ((y + obj_height / 2) < line_start)
{
(*dots_above) += 0.1;
*top_dot_x = (double) x / (double) (m_text->width);
}
else if ((y + obj_height / 2) > line_end)
{
(*dots_below) += 0.1;
}
else
{
(*dots_inside) += 0.1;
*inside_dot_x = (double) x / (double) (m_text->width);
}
}
else if (is_obj)
{
/* count */
if ((y + obj_height / 2) < line_start)
{
(*objs_above) += 0.1;
}
else if ((y + obj_height / 2) > line_end)
{
(*objs_below) += 0.1;
/* get metrics of bigest obj under font */
if (obj_width > *width)
{
obj_x = x;
obj_y = y;
obj_w = obj_width;
obj_h = obj_height;
}
}
else
{
(*objs_inside) += 0.1;
}
}
}
/* free obj map */
ho_objmap_free (o_obj);
/* if nothing below font we are finished */
if (!(*objs_below) && !(*dots_below))
return FALSE;
/* get nikud metrics */
*main_obj_height = 2.0 * (double) obj_h / (double) line_height;
*main_obj_width = 2.0 * (double) obj_w / (double) line_height;
if (obj_y && obj_h)
{
*top = 2.0 * (double) (obj_y - line_end) / (double) line_height;
*bottom =
2.0 * (double) ((obj_y + obj_h) - line_end) / (double) line_height;
}
/* get nikud start and end */
sum = 0;
for (y = line_end + 1; y < m_text->height && sum == 0; y++)
for (sum = 0, x = 1; x < (m_text->width - 1); x++)
sum += ho_bitmap_get (m_text, x, y);
font_start_y = y - 1;
sum = 0;
for (y = m_text->height - 1; y > font_start_y && sum == 0; y--)
for (sum = 0, x = 1; x < (m_text->width - 1); x++)
sum += ho_bitmap_get (m_text, x, y);
font_end_y = y + 1;
font_height = font_end_y - font_start_y;
if (!font_height)
return TRUE;
*height = 2.0 * (double) font_height / (double) line_height;
sum = 0;
for (x = 2; x < m_text->width && sum == 0; x++)
for (sum = 0, y = line_end; y < (m_text->height - 1); y++)
sum += ho_bitmap_get (m_text, x, y);
font_start_x = x - 1;
sum = 0;
for (sum = 0, x = m_text->width - 2; x > (font_start_x + 1) && sum == 0; x--)
for (sum = 0, y = line_end; y < (m_text->height - 1); y++)
sum += ho_bitmap_get (m_text, x, y);
font_end_x = x + 1;
font_width = font_end_x - font_start_x;
if (!font_width)
return TRUE;
*width = 2.0 * (double) font_width / (double) line_height;
/* get nikud egdes */
for (y = font_start_y, x = font_start_x;
x < font_end_x && y < (font_end_y)
&& !ho_bitmap_get (m_text, x, y); x++, y++) ;
*top_left = (double) (x - font_start_x) / (double) (line_height / 4);
if (*top_left > 1.0)
*top_left = 1.0;
for (y = font_end_y, x = font_start_x;
x < font_end_x && y > (font_start_y)
&& !ho_bitmap_get (m_text, x, y); x++, y--) ;
*bottom_left = (double) (x - font_start_x) / (double) (line_height / 4);
if (*bottom_left > 1.0)
*bottom_left = 1.0;
for (y = font_start_y, x = font_end_x - 1;
x > font_start_x && y < (font_end_y)
&& !ho_bitmap_get (m_text, x, y); x--, y++) ;
*top_right = (double) (font_end_x - x) / (double) (line_height / 4);
if (*top_right > 1.0)
*top_right = 1.0;
for (y = font_end_y, x = font_end_x - 1;
x > font_start_x && y > font_start_y
&& !ho_bitmap_get (m_text, x, y); x--, y--) ;
*bottom_right = (double) (font_end_x - x) / (double) (line_height / 4);
if (*bottom_right > 1.0)
*bottom_right = 1.0;
for (y = font_start_y, x = font_start_x + font_width / 2;
y < (font_start_y + font_height) && !ho_bitmap_get (m_text, x, y); y++) ;
*top_mid = (double) (y - font_start_y) / (double) (line_height / 4);
if (*top_mid > 1.0)
*top_mid = 1.0;
for (y = font_end_y, x = font_start_x + font_width / 2;
y > (font_start_y + font_height) && !ho_bitmap_get (m_text, x, y); y--) ;
*bottom_mid = (double) (font_end_y - y) / (double) (line_height / 4);
if (*bottom_mid > 1.0)
*bottom_mid = 1.0;
for (y = font_start_y + font_height / 2, x = font_start_x;
x < (font_end_x) && !ho_bitmap_get (m_text, x, y); x++) ;
*mid_left = (double) (x - font_start_x) / (double) (line_height / 4);
if (*mid_left > 1.0)
*mid_left = 1.0;
for (y = font_start_y + font_height / 2, x = font_end_x - 1;
x > (font_start_x) && !ho_bitmap_get (m_text, x, y); x--) ;
*mid_right = (double) (font_end_x - x) / (double) (line_height / 4);
if (*mid_right > 1.0)
*mid_right = 1.0;
return FALSE;
}
