개정판 ff956736
OCR이 같은 라인인지 판별하기 위한 Rect의 높이 수정
| DTI_PID/DTI_PID/tesseract_ocr_module.py | ||
|---|---|---|
| 28 | 28 |
|
| 29 | 29 |
''' |
| 30 | 30 |
@history 2018.04.26 Jeongwoo Make TextInfo object with Calculated Coords (with BoundBox Coords) |
| 31 |
2018.04.30 Jeongwoo Add QRect.setHeight() in if-statement [(lineRect is not None and currentRect is not None) and lineRect.intersects(currentRect)] |
|
| 31 | 32 |
''' |
| 32 | 33 |
def getTextInfo(img, startPoint, flag = FLAG_IMAGE_TO_BOXES, conf = """--psm 6 -c tessedit_char_whitelist=abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789-~.,/!@#$%&*(){}<>:;+=?\\"\\'"""):
|
| 33 | 34 |
im = Image.fromarray(img) |
| ... | ... | |
| 70 | 71 |
lastRT = (csx+cw, csy) |
| 71 | 72 |
lineSp = (csx, csy) |
| 72 | 73 |
lastCharHeight = charHeight |
| 73 |
lineRect = QRect(lineSp[0], lineSp[1], imgWidth - lineSp[0], ch) |
|
| 74 |
lineRect = QRect(lineSp[0], lineSp[1], imgWidth - lineSp[0], charHeight)
|
|
| 74 | 75 |
elif (lineRect is not None and currentRect is not None) and lineRect.intersects(currentRect): |
| 75 | 76 |
tempText = tempText + char |
| 76 | 77 |
lastRT = (csx+cw, min(lineSp[1], csy)) |
| 77 | 78 |
lineSp = (lineSp[0], min(lineSp[1], csy)) |
| 78 | 79 |
lastCharHeight = charHeight |
| 80 |
lineRect.setHeight(charHeight) |
|
| 79 | 81 |
else: |
| 80 | 82 |
# Save previous line |
| 81 | 83 |
prevLineText = ti.TextInfo(splitText[textGroupIndex], startPoint[0]+lineSp[0], startPoint[1]+lineSp[1], lastRT[0] - lineSp[0], lastCharHeight, 0) |
내보내기 Unified diff
