Spaces:
Sleeping
Sleeping
Update Counting_Columns_2_1.py
Browse files- Counting_Columns_2_1.py +64 -3
Counting_Columns_2_1.py
CHANGED
|
@@ -96,10 +96,37 @@ def get_columns_info(outsu4, img):
|
|
| 96 |
|
| 97 |
def getTextsPoints(x):
|
| 98 |
point_list = []
|
|
|
|
| 99 |
for h in x:
|
| 100 |
-
point_list.append((h[
|
| 101 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 102 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 103 |
|
| 104 |
def distance(point1, point2):
|
| 105 |
x1, y1 = point1
|
|
@@ -139,6 +166,40 @@ def generate_legend(found_tuple):
|
|
| 139 |
return df
|
| 140 |
|
| 141 |
def mainfun(plan):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 142 |
texts_from_pdf = get_text_from_pdf(plan)
|
| 143 |
img = convert2img(plan)
|
| 144 |
imgResult = segment_brown(img)
|
|
@@ -166,4 +227,4 @@ def mainfun(plan):
|
|
| 166 |
nearby = fix_rotation_90(pc_coordinates)
|
| 167 |
columns_types = getColumnsTypes(nearby, texts_from_pdf)
|
| 168 |
legend = generate_legend(columns_types)
|
| 169 |
-
return legend
|
|
|
|
| 96 |
|
| 97 |
def getTextsPoints(x):
|
| 98 |
point_list = []
|
| 99 |
+
pt_clm = {}
|
| 100 |
for h in x:
|
| 101 |
+
point_list.append(calculate_midpoint(h[1],h[0],h[3],h[2]))
|
| 102 |
+
pt_clm[calculate_midpoint(h[1],h[0],h[3],h[2])] = h[4]
|
| 103 |
+
return point_list, pt_clm
|
| 104 |
+
|
| 105 |
+
def fix_90_ky_val(pt_clm):
|
| 106 |
+
new_derotated = {}
|
| 107 |
+
for ky in pt_clm:
|
| 108 |
+
pts = fitz.Point(ky[0], ky[1]) * derotationMatrix
|
| 109 |
+
new_ky = ((int(pts.y),int(pts.x)))
|
| 110 |
+
new_derotated[new_ky] = pt_clm[ky]
|
| 111 |
+
return new_derotated
|
| 112 |
+
|
| 113 |
+
def calculate_midpoint(x1,y1,x2,y2):
|
| 114 |
+
xm = int((x1 + x2) / 2)
|
| 115 |
+
ym = int((y1 + y2) / 2)
|
| 116 |
+
return (xm, ym)
|
| 117 |
+
|
| 118 |
+
def getColumnsTypesKeyValue(nearbyy, pt_clm):
|
| 119 |
+
words = []
|
| 120 |
+
for i in range(len(nearbyy)):
|
| 121 |
+
words.append(pt_clm[nearbyy[i]])
|
| 122 |
+
return words
|
| 123 |
|
| 124 |
+
def fix_rotation_90(pc_coordinates, derotationMatrix):
|
| 125 |
+
coor = []
|
| 126 |
+
for coordinate in pc_coordinates:
|
| 127 |
+
pts = fitz.Point(coordinate[0], coordinate[1]) * derotationMatrix
|
| 128 |
+
coor.append((int(pts.y),int(pts.x)))
|
| 129 |
+
return coor
|
| 130 |
|
| 131 |
def distance(point1, point2):
|
| 132 |
x1, y1 = point1
|
|
|
|
| 166 |
return df
|
| 167 |
|
| 168 |
def mainfun(plan):
|
| 169 |
+
pdf_document = fitz.open(plan)
|
| 170 |
+
page = pdf_document[0]
|
| 171 |
+
rotation = page.rotation
|
| 172 |
+
derotationMatrix=page.derotation_matrix
|
| 173 |
+
texts_from_pdf = get_text_from_pdf(plan)
|
| 174 |
+
text_points, txtpts_ky_vlu = getTextsPoints(texts_from_pdf)
|
| 175 |
+
if rotation != 0:
|
| 176 |
+
if rotation ==90:
|
| 177 |
+
text_points = fix_rotation_90(text_points, derotationMatrix)
|
| 178 |
+
txtpts_ky_vlu = fix_90_ky_val(txtpts_ky_vlu)
|
| 179 |
+
|
| 180 |
+
img = convert2img(plan)
|
| 181 |
+
imgResult = segment_brown(img)
|
| 182 |
+
outsu = threshold(imgResult)
|
| 183 |
+
column_points,mask_clmns, mask_walls = get_columns_info(outsu, img)
|
| 184 |
+
|
| 185 |
+
if len(column_points) > 10:
|
| 186 |
+
# BROWN COLUMNS
|
| 187 |
+
nearby = getNearestText(text_points, column_points)
|
| 188 |
+
columns_types_v = getColumnsTypesKeyValue(nearby, txtpts_ky_vlu)
|
| 189 |
+
legend = generate_legend(columns_types_v)
|
| 190 |
+
|
| 191 |
+
else:
|
| 192 |
+
# BLUE COLUMNS
|
| 193 |
+
img_blue = changeGrayModify(img)
|
| 194 |
+
imgResult = segment_blue(img_blue)
|
| 195 |
+
outsu = threshold(imgResult)
|
| 196 |
+
column_points,mask_clmns, mask_walls = get_columns_info(outsu, img)
|
| 197 |
+
nearby = getNearestText(text_points, column_points)
|
| 198 |
+
columns_types_v = getColumnsTypesKeyValue(nearby, txtpts_ky_vlu)
|
| 199 |
+
legend = generate_legend(columns_types_v)
|
| 200 |
+
return legend
|
| 201 |
+
|
| 202 |
+
'''def mainfun(plan):
|
| 203 |
texts_from_pdf = get_text_from_pdf(plan)
|
| 204 |
img = convert2img(plan)
|
| 205 |
imgResult = segment_brown(img)
|
|
|
|
| 227 |
nearby = fix_rotation_90(pc_coordinates)
|
| 228 |
columns_types = getColumnsTypes(nearby, texts_from_pdf)
|
| 229 |
legend = generate_legend(columns_types)
|
| 230 |
+
return legend'''
|