def DetectVATInvoice(box, symbol, filePath): # 识别增值税发票种类1
# 定位二维码
# 定位×
dic = xmlToDict.XmlTodict('ModeLabel_00001.xml')
tplt = [
dic['QRCode'][0], dic['QRCode'][1], dic['figureX'][2],
dic['figureX'][3]
]
for c in tplt:
if c == None:
print('Templet VATInvoice error')
for item in dic:
if item != 'QRCode' and item != 'figureX':
# print(item)
tmp = MakeFileInV([[int(dic.get(item)[0]),
int(dic.get(item)[1])],
[int(dic.get(item)[2]),
int(dic.get(item)[3])]], box, symbol, filePath,
item, tplt)
print(item + ": " + OcrPic(tmp))
js = InterfaceType.JsonInterface.invoice()
js.addVATInvoiceInfo
def mubanDetectInvoiceDate(filepath, setKey='invoiceDate'):
midProcessResult = [None, None]
midProcessResult[0] = filepath
midProcessResult[1] = 11
# vat发票专票
VATInvoiceTemplet = {
}
dic = xmlToDict.XmlTodict('/home/huangzheng/ocr/VATInvoiceSimpleMuban.xml')
# tplt = [dic['QRCode'][0], dic['QRCode'][1], dic['figureX'][0] + dic['figureX'][2] / 2, dic['figureX'][1] + dic['figureX'][3] / 2]
tplt = [dic['figureX'][0] + dic['figureX'][2] / 2, dic['figureX'][1] + dic['figureX'][3] / 2]
# print(tplt)
'''
for c in tplt:
if c == None:
print('Templet VATInvoice error')
'''
TemType = {}
if midProcessResult[1] == 11: # 增值税专用
VATInvoiceTemplet[setKey] = [int(dic.get(setKey)[0]), int(dic.get(setKey)[1]), int(dic.get(setKey)[2]),
int(dic.get(setKey)[3])]
TemType = VATInvoiceTemplet
fcv = cv2.imread(filepath, 1)
# print(fcv)
try:
w1 = fcv.shape
except:
print("picture is None")
if w1[0] + w1[1] > 1500:
rate = 0.5
# print("rate : 0.5")
if midProcessResult[1] == 11:
# box = Detect.detect(cv2.imread(midProcessResult[0]), rate)
figureP = FindCircle.findSymbol(filepath)
# StBox = sortBox(box)
# print(box)
# print(figureP)
# print(StBox)
if figureP == None:
return None
Templet = simplyAdjust(TemType, [figureP[0], figureP[1]], tplt, w1) # 增值税专票
attributeLine = lineToAttribute.getAtbt.compute(textline(midProcessResult[0]), Templet)
return attributeLine
def init(filepath, type):
# filepath = 'Image_065.jpg'
'''if len(sys.argv) > 1:
filePathg = sys.argv[1]
filepath = filePathg
'''
# print(sys.argv)
# print(filepath)
if type == None:
midProcessResult = detectType.detectType(filepath) # tangpeng 预处理
# 未分类
else:
if type == 'blue' or type == 'excess':
midProcessResult = PipeInvoice.getPipe(filepath, type, False)
else:
##type == 'red'
midProcessResult = [None, None, None]
midProcessResult[0] = filepath
midProcessResult[1] = 2 # 专票
midProcessResult[2] = textline(midProcessResult[0])
# midProcessResult = [None, None]
# midProcessResult[0] = midPR[0]
# midProcessResult[1] = midPR[1]
# midProcessResult[0] = filepath
# 行提取
blueTemplet = {
'departCity': [48, 62, 222, 56],
'arriveCity': [412, 61, 228, 55],
'trainNumber': [264, 62, 170, 57],
'invoiceDate': [24, 139, 369, 42],
'seatNum': [408, 138, 160, 40],
'idNum': [22, 276, 306, 38],
'passenger': [328, 276, 150, 38],
'totalAmount': [33, 177, 151, 39],
'ticketsNum': [21, 10, 195, 66]
}
''' 'departCity': [29, 74, 218, 54],
'arriveCity': [425, 68, 224, 64],
'trainNumber': [230, 65, 203, 62],
'invoiceDate': [0, 163, 357, 41],
'seatNum': [392, 164, 203, 46],
'idNum': [0, 343, 350, 45],
'totalAmount': [3, 206, 212, 52],
'ticketsNum': [34, 40, 202, 47]
}'''
redTemplet = {
'idNum': [66, 242, 357, 38],
'departCity': [66, 66, 222, 45],
'arriveCity': [388, 66, 225, 47],
'trainNumber': [288, 53, 103, 41],
'invoiceDate': [66, 114, 237, 43],
'seatNum': [400, 115, 210, 46],
'totalAmount': [66, 163, 188, 34],
'ticketsNum': [21, 23, 218, 47]
}
''''''
excessTemplet = {
'departCity': [26, 40, 151, 33],
'arriveCity': [271, 40, 169, 33],
'trainNumber': [178, 35, 92, 32],
'invoiceDate': [12, 82, 203, 29],
'seatNum': [315, 79, 136, 33],
'totalAmount': [12, 118, 167, 28],
'ticketsNum': [23, 12, 147, 37]
}
# vatinvoice
VATInvoiceTemplet = {}
dic = xmlToDict.XmlTodict('VATInvoiceMuban.xml')
tplt = [
dic['QRCode'][0], dic['QRCode'][1],
dic['figureX'][0] + dic['figureX'][2] / 2,
dic['figureX'][1] + dic['figureX'][3] / 2
]
# print(tplt)
for c in tplt:
if c == None:
print('Templet VATInvoice error')
# print(VATInvoiceTemplet)
# midProcessResult[1] = 2#专票 for test
# TemType = redTemplet #默认蓝票
if midProcessResult[1] == 1:
TemType = blueTemplet
if midProcessResult[1] == 2:
TemType = redTemplet
if midProcessResult[1] == 3:
TemType = excessTemplet
if midProcessResult[1] == 11: # 增值税专用 预留
for item in dic:
if item != 'QRCode' and item != 'figureX':
# print(item)
# tmp = MakeFileInV([[int(dic.get(item)[0]), int(dic.get(item)[1])], [int(dic.get(item)[2]), int(dic.get(item)[3])]], box, symbol, filePath, item, tplt)
VATInvoiceTemplet[item] = [
int(dic.get(item)[0]),
int(dic.get(item)[1]),
int(dic.get(item)[2]),
int(dic.get(item)[3])
]
TemType = VATInvoiceTemplet
rate = 1
# print(filepath)
fcv = cv2.imread(filepath, 1)
# print(fcv)
w1 = fcv.shape
if w1[0] + w1[1] > 1500:
rate = 0.5
print("rate : 0.5")
if midProcessResult[1] == 1:
box = Detect.detect(cv2.imread(midProcessResult[0]), rate)
Templet = adjustToTextLine(TemType, box, midProcessResult[1],
None) # 蓝火车票
if midProcessResult[1] == 2:
rate = 2.0
print("rate : 2.0")
box = Detect.detect(cv2.imread(midProcessResult[0]), rate)
# print( box.tolist())
Templet = adjustToTextLine(TemType, box, midProcessResult[1],
None) # 红火车票
if midProcessResult[1] == 3:
rate = 1.0
print("rate : 1.0")
box = Detect.detect(cv2.imread(midProcessResult[0]), rate)
# print( box.tolist())
Templet = adjustToTextLine(TemType, box, midProcessResult[1],
None) # 红(补票)车票
if midProcessResult[1] == 11:
box = Detect.detect(cv2.imread(midProcessResult[0]), rate)
figureP = FindCircle.findSymbol(filepath)
StBox = sortBox(box)
# print(box)
# print(figureP)
# print(StBox)
Templet = adjustToTextLine(
TemType, [StBox[0], StBox[1], figureP[0], figureP[1]],
midProcessResult[1], tplt) # 增值税专票
im = cv2.imread(filepath, 0)
rec = []
for c in TemType:
rec.append(TemType[c])
vis_textline0 = fp.util.visualize.rects(im, rec)
# vis_textline1 = fp.util.visualize.rects(im, rects, types)
# 显示
pl.figure(figsize=(15, 10))
pl.subplot(2, 2, 1)
pl.imshow(im, 'gray')
pl.subplot(2, 2, 2)
pl.imshow(vis_textline0)
pl.show()
# print(textline(midProcessResult[0]))
attributeLine = lineToAttribute.getAtbt.compute(midProcessResult[2],
Templet)
# print(attributeLine)
# print(type(attributeLine))
# print(attributeLine['departCity'])
jsonResult = flow.cropToOcr(midProcessResult[0], attributeLine,
midProcessResult[1]) # ocr和分词
print(jsonResult)
return jsonResult
def mubanDetectInvoiceDate(filepath, setKey='invoiceDate'):
midProcessResult = [None, None]
midProcessResult[0] = filepath
midProcessResult[1] = 11
# vat发票专票
VATInvoiceTemplet = {}
dic = xmlToDict.XmlTodict('/home/huangzheng/ocr/VATInvoiceSimpleMuban.xml')
# tplt = [dic['QRCode'][0], dic['QRCode'][1], dic['figureX'][0] + dic['figureX'][2] / 2, dic['figureX'][1] + dic['figureX'][3] / 2]
tplt = [
dic['figureX'][0] + dic['figureX'][2] / 2,
dic['figureX'][1] + dic['figureX'][3] / 2
]
# print(tplt)
'''
for c in tplt:
if c == None:
print('Templet VATInvoice error')
'''
TemType = {}
if midProcessResult[1] == 11: # 增值税专用
VATInvoiceTemplet[setKey] = [
int(dic.get(setKey)[0]),
int(dic.get(setKey)[1]),
int(dic.get(setKey)[2]),
int(dic.get(setKey)[3])
]
TemType = VATInvoiceTemplet
fcv = cv2.imread(filepath, 1)
# print(fcv)
try:
w1 = fcv.shape
except:
print("picture is None")
if w1[0] + w1[1] > 1500:
rate = 0.5
# print("rate : 0.5")
if midProcessResult[1] == 11:
# box = Detect.detect(cv2.imread(midProcessResult[0]), rate)
figureP = FindCircle.findSymbol(filepath)
# StBox = sortBox(box)
# print(box)
# print(figureP)
# print(StBox)
if figureP == None:
return None
Templet = simplyAdjust(TemType, [figureP[0], figureP[1]], tplt,
w1) # 增值税专票
attributeLine = lineToAttribute.getAtbt.compute(
textline(midProcessResult[0]), Templet)
# 生成行提取的图片
plt_rects = []
for x in attributeLine:
plt_rects.append(attributeLine[x])
# 显示
vis_textline0 = fp.util.visualize.rects(
cv2.imread(midProcessResult[0], 0), plt_rects)
pl.imshow(vis_textline0)
# 保存到line目录
pltpath = midProcessResult[0].replace("upload", "line")
try:
pl.savefig(pltpath)
except Exception as e:
print("绘制行提取图片不支持bmp格式：{}".format(e))
return attributeLine
def surface(filename, type='blue'):
filepath = os.path.join('allstatic', filename)
# 原方法
if type == None:
midProcessResult = detectType.detectType('allstatic', filename) # tangpeng 预处理
# 未分类
else:
if type == 'blue' or type == 'excess':
# return out_filename, 1, pipe.textlines, cdic
midProcessResult = PipeInvoice.getPipe('allstatic', filename, type, False)
else:
##type == 'red'
midProcessResult = [None, None, None]
out_filename = filename.replace('upload', 'out')
out_filename = os.path.join('allstatic', out_filename)
# 拷贝到out
shutil.copy(filepath, out_filename)
midProcessResult[0] = out_filename
midProcessResult[1] = 2 # 专票
# 暂时用原图，不用校正后的图
midProcessResult[2] = textline(midProcessResult[0])
# 行提取
blueTemplet = {
'departCity': [48, 62, 222, 56],
'arriveCity': [412, 61, 228, 55],
'trainNumber': [264, 62, 170, 57],
'invoiceDate': [24, 139, 369, 42],
'seatNum': [408, 138, 160, 40],
'idNum': [22, 276, 306, 38],
'passenger': [328, 276, 150, 38],
'totalAmount': [33, 177, 151, 39],
'ticketsNum': [21, 10, 195, 66]
}
''' 'departCity': [29, 74, 218, 54],
'arriveCity': [425, 68, 224, 64],
'trainNumber': [230, 65, 203, 62],
'invoiceDate': [0, 163, 357, 41],
'seatNum': [392, 164, 203, 46],
'idNum': [0, 343, 350, 45],
'totalAmount': [3, 206, 212, 52],
'ticketsNum': [34, 40, 202, 47]
}'''
redTemplet = {
'idNum': [66, 242, 357, 38],
'departCity': [66, 66, 222, 45],
'arriveCity': [388, 66, 225, 47],
'trainNumber': [288, 53, 103, 41],
'invoiceDate': [66, 114, 237, 43],
'seatNum': [400, 115, 210, 46],
'totalAmount': [66, 163, 188, 34],
'ticketsNum': [21, 23, 218, 47]
}
''''''
excessTemplet = {
'departCity': [26, 40, 151, 33],
'arriveCity': [271, 40, 169, 33],
'trainNumber': [178, 35, 92, 32],
'invoiceDate': [12, 82, 203, 29],
'seatNum': [315, 79, 136, 33],
'totalAmount': [12, 118, 167, 28],
'ticketsNum': [23, 12, 147, 37]
}
# vatinvoice
VATInvoiceTemplet = {
}
if midProcessResult[1] == 1:
TemType = blueTemplet
if midProcessResult[1] == 2:
TemType = redTemplet
if midProcessResult[1] == 3:
TemType = excessTemplet
if midProcessResult[1] == 11: # 增值税专用 预留
dic = xmlToDict.XmlTodict('VATInvoiceMuban.xml')
tplt = [dic['QRCode'][0], dic['QRCode'][1], dic['figureX'][0] + dic['figureX'][2] / 2,
dic['figureX'][1] + dic['figureX'][3] / 2]
# print(tplt)
for c in tplt:
if c == None:
print('Templet VATInvoice error')
for item in dic:
if item != 'QRCode' and item != 'figureX':
# print(item)
# tmp = MakeFileInV([[int(dic.get(item)[0]), int(dic.get(item)[1])], [int(dic.get(item)[2]), int(dic.get(item)[3])]], box, symbol, filePath, item, tplt)
VATInvoiceTemplet[item] = [int(dic.get(item)[0]), int(dic.get(item)[1]), int(dic.get(item)[2]),
int(dic.get(item)[3])]
TemType = VATInvoiceTemplet
rate = 1
fcv = cv2.imread(filepath, 1)
w1 = fcv.shape
if w1[0] + w1[1] > 1500:
rate = 0.5
# print("rate : 0.5")
if midProcessResult[1] == 1:
box = Detect.detect(cv2.imread(midProcessResult[0]), rate)
Templet = adjustToTextLine(TemType, box, midProcessResult[1], None) # 蓝火车票
if midProcessResult[1] == 2:
rate = 2.0
print("rate : 2.0")
box = Detect.detect(cv2.imread(midProcessResult[0]), rate)
# print( box.tolist())
Templet = adjustToTextLine(TemType, box, midProcessResult[1], None) # 红火车票
if midProcessResult[1] == 3:
rate = 1.0
print("rate : 1.0")
box = Detect.detect(cv2.imread(midProcessResult[0]), rate)
# print( box.tolist())
Templet = adjustToTextLine(TemType, box, midProcessResult[1], None) # 红(补票)车票
if midProcessResult[1] == 11:
box = Detect.detect(cv2.imread(midProcessResult[0]), rate)
figureP = FindCircle.findSymbol(filepath)
StBox = sortBox(box)
Templet = adjustToTextLine(TemType, [StBox[0], StBox[1], figureP[0], figureP[1]], midProcessResult[1],
tplt) # 增值税专票
attributeLine = lineToAttribute.getAtbt.compute(midProcessResult[2], Templet)
# 生成行提取的图片
plt_rects = []
for x in attributeLine:
plt_rects.append(attributeLine[x])
# 显示
vis_textline0 = fp.util.visualize.rects(cv2.imread(midProcessResult[0], 0), plt_rects)
pl.imshow(vis_textline0)
# 保存到line目录
pltpath = midProcessResult[0].replace("out", "line")
try:
pl.savefig(pltpath)
except Exception as e:
print("绘制行提取图片不支持bmp格式：{}".format(e))
pass
return midProcessResult[0], midProcessResult[1], attributeLine