1.ImageMagick
2.tesseract-OCR
3.Python3.7
4.from PIL import Image as PI
5.import io
6.import os
7.import pyocr.builders
8.from cnocr import CnOcr
import データはデジタル形式に変換されますが、これに基づいて、最初に大文字の漢字と数字の変換を完了する必要があります。
def chineseNumber2Int(strNum: str): result = 0 temp = 1 # 存放一个单位的数字如:十万 count = 0 # 判断是否有chArr cnArr = ['壹', '贰', '叁', '肆', '伍', '陆', '柒', '捌', '玖'] chArr = ['拾', '佰', '仟', '万', '亿'] for i in range(len(strNum)): b = True c = strNum[i] for j in range(len(cnArr)): if c == cnArr[j]: if count != 0: result += temp count = 0 temp = j + 1 b = False break if b: for j in range(len(chArr)): if c == chArr[j]: if j == 0: temp *= 10 elif j == 1: temp *= 100 elif j == 2: temp *= 1000 elif j == 3: temp *= 10000 elif j == 4: temp *= 100000000 count += 1 if i == len(strNum) - 1: result += temp return result
上記のコードは、大文字と数字を変換するために使用できます。たとえば、「200000」をエクスポートするには「2万元」と入力し、それを数字に変換すると、表が大幅に簡素化されます。テーブル操作を完了しながらデータをアーカイブするのに役立ちます。
次に、請求書の内部内容を分析する必要がありますが、次の図から、「請求書の発行日」、「為替手形の到着日」のデータ内容を取得する必要があることがわかります。 , 「請求書番号」「受取人」「請求金額」「引き出し」を描画ソフトで正確に配置できます。図に示すように、小さな黒い点がマウスの位置で、描画ソフトウェアの左下隅がその座標です。
発行日を抽出します
def text1(new_img): #提取出票日期 left = 80 top = 143 right = 162 bottom = 162 image_text1 = new_img.crop((left, top, right, bottom)) #展示图片 #image_text1.show() txt1 = tool.image_to_string(image_text1) print(txt1) return str(txt1)
金額を引き出します
def text2(new_img): #提取金额 left = 224 top = 355 right = 585 bottom = 380 image_text2 = new_img.crop((left, top, right, bottom)) #展示图片 #image_text2.show() image_text2.save("img/tmp.png") temp = ocr.ocr("img/tmp.png") temp="".join(temp[0]) txt2=chineseNumber2Int(temp) print(txt2) return txt2
def text3(new_img): #提取出票人 left = 177 top = 207 right = 506 bottom = 231 image_text3 = new_img.crop((left, top, right, bottom)) #展示图片 #image_text3.show() image_text3.save("img/tmp.png") temp = ocr.ocr("img/tmp.png") txt3="".join(temp[0]) print(txt3) return txt3
def text4(new_img): #提取付款行 left = 177 top = 274 right = 492 bottom = 311 image_text4 = new_img.crop((left, top, right, bottom)) #展示图片 #image_text4.show() image_text4.save("img/tmp.png") temp = ocr.ocr("img/tmp.png") txt4="".join(temp[0]) print(txt4) return txt4
def text5(new_img): #提取汇票到日期 left = 92 top = 166 right = 176 bottom = 184 image_text5 = new_img.crop((left, top, right, bottom)) #展示图片 #image_text5.show() txt5 = tool.image_to_string(image_text5) print(txt5) return txt5
def text6(new_img): #提取票据号码 left = 598 top = 166 right = 870 bottom = 182 image_text6 = new_img.crop((left, top, right, bottom)) #展示图片 #image_text6.show() txt6 = tool.image_to_string(image_text6) print(txt6) return txt6
ocr=CnOcr() tool = pyocr.get_available_tools()[0] filePath='img' img_name=[] for i,j,name in os.walk(filePath): img_name=name
count=1 book = xlwt.Workbook(encoding='utf-8',style_compression=0) sheet = book.add_sheet('test',cell_overwrite_ok=True) for i in img_name: img_url = filePath+"/"+i with open(img_url, 'rb') as f: a = f.read() new_img = PI.open(io.BytesIO(a)) ## 写入csv col = ('年份','出票日期','金额','出票人','付款行全称','汇票到日期','备注') for j in range(0,7): sheet.write(0,j,col[j]) book.save('1.csv') shijian=text1(new_img) sheet.write(count,0,shijian[0:4]) sheet.write(count,1,shijian[5:]) sheet.write(count,2,text2(new_img)) sheet.write(count,3,text3(new_img)) sheet.write(count,4,text4(new_img)) sheet.write(count,5,text5(new_img)) sheet.write(count,6,text6(new_img)) count = count + 1
from wand.image import Image from PIL import Image as PI import pyocr import io import re import os import shutil import pyocr.builders from cnocr import CnOcr import requests import xlrd import xlwt from openpyxl import load_workbook def chineseNumber2Int(strNum: str): result = 0 temp = 1 # 存放一个单位的数字如:十万 count = 0 # 判断是否有chArr cnArr = ['壹', '贰', '叁', '肆', '伍', '陆', '柒', '捌', '玖'] chArr = ['拾', '佰', '仟', '万', '亿'] for i in range(len(strNum)): b = True c = strNum[i] for j in range(len(cnArr)): if c == cnArr[j]: if count != 0: result += temp count = 0 temp = j + 1 b = False break if b: for j in range(len(chArr)): if c == chArr[j]: if j == 0: temp *= 10 elif j == 1: temp *= 100 elif j == 2: temp *= 1000 elif j == 3: temp *= 10000 elif j == 4: temp *= 100000000 count += 1 if i == len(strNum) - 1: result += temp return result def text1(new_img): #提取出票日期 left = 80 top = 143 right = 162 bottom = 162 image_text1 = new_img.crop((left, top, right, bottom)) #展示图片 #image_text1.show() txt1 = tool.image_to_string(image_text1) print(txt1) return str(txt1) def text2(new_img): #提取金额 left = 224 top = 355 right = 585 bottom = 380 image_text2 = new_img.crop((left, top, right, bottom)) #展示图片 #image_text2.show() image_text2.save("img/tmp.png") temp = ocr.ocr("img/tmp.png") temp="".join(temp[0]) txt2=chineseNumber2Int(temp) print(txt2) return txt2 def text3(new_img): #提取出票人 left = 177 top = 207 right = 506 bottom = 231 image_text3 = new_img.crop((left, top, right, bottom)) #展示图片 #image_text3.show() image_text3.save("img/tmp.png") temp = ocr.ocr("img/tmp.png") txt3="".join(temp[0]) print(txt3) return txt3 def text4(new_img): #提取付款行 left = 177 top = 274 right = 492 bottom = 311 image_text4 = new_img.crop((left, top, right, bottom)) #展示图片 #image_text4.show() image_text4.save("img/tmp.png") temp = ocr.ocr("img/tmp.png") txt4="".join(temp[0]) print(txt4) return txt4 def text5(new_img): #提取汇票到日期 left = 92 top = 166 right = 176 bottom = 184 image_text5 = new_img.crop((left, top, right, bottom)) #展示图片 #image_text5.show() txt5 = tool.image_to_string(image_text5) print(txt5) return txt5 def text6(new_img): #提取票据号码 left = 598 top = 166 right = 870 bottom = 182 image_text6 = new_img.crop((left, top, right, bottom)) #展示图片 #image_text6.show() txt6 = tool.image_to_string(image_text6) print(txt6) return txt6 ocr=CnOcr() tool = pyocr.get_available_tools()[0] filePath='img' img_name=[] for i,j,name in os.walk(filePath): img_name=name count=1 book = xlwt.Workbook(encoding='utf-8',style_compression=0) sheet = book.add_sheet('test',cell_overwrite_ok=True) for i in img_name: img_url = filePath+"/"+i with open(img_url, 'rb') as f: a = f.read() new_img = PI.open(io.BytesIO(a)) ## 写入csv col = ('年份','出票日期','金额','出票人','付款行全称','汇票到日期','备注') for j in range(0,7): sheet.write(0,j,col[j]) book.save('1.csv') shijian=text1(new_img) sheet.write(count,0,shijian[0:4]) sheet.write(count,1,shijian[5:]) sheet.write(count,2,text2(new_img)) sheet.write(count,3,text3(new_img)) sheet.write(count,4,text4(new_img)) sheet.write(count,5,text5(new_img)) sheet.write(count,6,text6(new_img)) count = count + 1
以上がPython を使用してバッチデータ抽出を実装する方法の詳細内容です。詳細については、PHP 中国語 Web サイトの他の関連記事を参照してください。