개정판 3ed85f79
dev issue #507 : fix text
Change-Id: If3c96639dbc60f9ae3ad11da2525749e8449b407
| DTI_PID/DTI_PID/tesseract_ocr_module.py | ||
|---|---|---|
| 103 | 103 |
boundaryOcrData = pytesseract.image_to_boxes(im, config=_conf, lang=oCRLang) |
| 104 | 104 |
bounding_boxes = boundaryOcrData.split('\n')
|
| 105 | 105 |
merged_boxes = [] |
| 106 |
|
|
| 107 |
# before |
|
| 108 |
""" |
|
| 106 | 109 |
for box in bounding_boxes: |
| 107 | 110 |
if merged_boxes: |
| 108 | 111 |
tokens = box.split(' ')
|
| ... | ... | |
| 130 | 133 |
maxx = int(tokens[3]) |
| 131 | 134 |
maxy = int(tokens[4]) |
| 132 | 135 |
merged_boxes.append(QRect(minx, miny, maxx - minx, maxy - miny)) |
| 133 |
|
|
| 136 |
""" |
|
| 137 |
|
|
| 138 |
for box in bounding_boxes: |
|
| 139 |
if merged_boxes: |
|
| 140 |
tokens = box.split(' ')
|
|
| 141 |
if len(tokens) >= 5: |
|
| 142 |
minx = int(tokens[1]) |
|
| 143 |
miny = int(tokens[2]) |
|
| 144 |
maxx = int(tokens[3]) |
|
| 145 |
maxy = int(tokens[4]) |
|
| 146 |
|
|
| 147 |
findBox = None |
|
| 148 |
for merged_box in merged_boxes: |
|
| 149 |
bottom = merged_box.bottom() |
|
| 150 |
top = merged_box.top() |
|
| 151 |
if abs(miny - top) <= 1 or abs(maxy - bottom) <= 1: |
|
| 152 |
findBox = merged_box |
|
| 153 |
|
|
| 154 |
if findBox: |
|
| 155 |
findBox.setLeft(min(findBox.left(), minx)) |
|
| 156 |
findBox.setTop(min(findBox.top(), miny)) |
|
| 157 |
findBox.setRight(max(findBox.right(), maxx)) |
|
| 158 |
findBox.setBottom(max(findBox.bottom(), maxy)) |
|
| 159 |
else: |
|
| 160 |
merged_boxes.append(QRect(minx, miny, maxx - minx, maxy - miny)) |
|
| 161 |
|
|
| 162 |
else: |
|
| 163 |
tokens = box.split(' ')
|
|
| 164 |
if len(tokens) >= 5: |
|
| 165 |
minx = int(tokens[1]) |
|
| 166 |
miny = int(tokens[2]) |
|
| 167 |
maxx = int(tokens[3]) |
|
| 168 |
maxy = int(tokens[4]) |
|
| 169 |
merged_boxes.append(QRect(minx, miny, maxx - minx, maxy - miny)) |
|
| 170 |
|
|
| 134 | 171 |
for rect in merged_boxes: |
| 135 | 172 |
if not rect.isValid() or \ |
| 136 | 173 |
rect.left() < 0 or rect.top() < 0 or rect.right() > imgWidth or rect.bottom() > imgHeight: continue |
내보내기 Unified diff
