def ocr2PDF(ocrFile):
zoomRatio=10
p, lines, words=readOCR(ocrFile)
c=canvas.Canvas('static/test.pdf', bottomup=0,pagesize=(p.right/zoomRatio,p.bottom/zoomRatio))
# font=ImageFont.truetype('TimesNewRoman', size=10)
# font=ImageFont.load('Helvetica')
for l in lines:
for w in l.words:
textWidth = stringWidth(w.text, 'Helvetica', 10)
print w.text
if textWidth==0:
continue
fontSize=round(10.0*(w.right-w.left)/zoomRatio/textWidth)
c.setFont('Helvetica',fontSize)
c.drawString(w.left/zoomRatio,l.bottom/zoomRatio, w.text)
c.save()
def refine_process():
lang = "eng"
image = Image.open(workFile)
pages, lines, words = readOCR(outputFile + ".html")
resp = ""
for n in xrange(len(lines)):
l = lines[n]
tmpImage = "static/tmp/lineImage" + str(n) + ".png"
tmpXML = "static/tmp/lineImage" + str(n)
lineImage = image.crop((l.left, l.top, l.right, l.bottom)).convert("RGB")
w = l.right - l.left
h = l.bottom - l.top
scale = 1
lineImage = lineImage.resize((int(w * scale), int(h * scale)), Image.ANTIALIAS)
lineImage.filter(ImageFilter.SMOOTH).filter(ImageFilter.BLUR)
ImageEnhance.Contrast(lineImage)
margin = 500
boxImage = Image.new("L", (lineImage.size[0] + margin, lineImage.size[1] + margin), "white")
boxImage.paste(lineImage, (margin / 2, margin / 2))
# boxImage=boxImage.convert('L')
# boxImage=boxImage.point(lambda i: i>180 and 255)
# boxImage=boxImage.resize((boxImage.size[0],boxImage.size[1]), Image.ANTIALIAS)
# boxImage=boxImage.filter(ImageFilter.SHARPEN)
boxImage.save(tmpImage)
command = ["tesseract", tmpImage, tmpXML, "-l", lang, "hocr"]
proc = sp.Popen(command, stderr=sp.PIPE)
proc.wait()
linePage, lineLines, lineWords = readOCR(tmpXML + ".html")
for ll in lineLines:
for w in ll.words:
resp += w.text + " "
resp += "\n"
return resp
def process_image(image, lang="eng"):
image.save(inputFile)
image = pre_process(image).convert("RGB")
image.filter(ImageFilter.SMOOTH)
image.save(workFile)
crude_process(lang=lang)
pages, lines, words = readOCR(outputFile + ".html")
resp = ""
for l in lines:
# resp+='<p>'
for w in l.words:
resp += w.text + " "
# resp+='</p>'
resp += "\n"
return resp
return resp
