基于Python与OCR的票据图片批量数据提取实战

def chineseNumber2Int(strNum: str):
    result = 0
    temp = 1  # 存放一个单位的基数，如十万
    count = 0  # 记录连续出现的单位数
    cnArr = ['壹', '贰', '叁', '肆', '伍', '陆', '柒', '捌', '玖']
    chArr = ['拾', '佰', '仟', '万', '亿']
    
    for char in strNum:
        base_found = False
        for val_idx, digit in enumerate(cnArr):
            if char == digit:
                if count != 0:
                    result += temp
                    count = 0
                temp = val_idx + 1
                base_found = True
                break
        if not base_found:
            for unit_idx, unit in enumerate(chArr):
                if char == unit:
                    multipliers = [10, 100, 1000, 10000, 100000000]
                    temp *= multipliers[unit_idx]
                count += 1
    
    result += temp
    return result

def extract_date(new_img, tool):
    # 出票日期区域
    cropped = new_img.crop((80, 143, 162, 162))
    return str(tool.image_to_string(cropped))

def extract_amount(new_img, ocr_engine):
    # 金额区域
    cropped = new_img.crop((224, 355, 585, 380))
    raw_text = "".join(ocr_engine.ocr("temp_crop.png")[0])
    return chineseNumber2Int(raw_text)

import os
from PIL import Image as PI
import io
from cnocr import CnOcr
import pyocr
import xlwt

# 1. 初始化引擎
ocr_cn = CnOcr()
tool_tess = pyocr.get_available_tools()[0]

file_dir = 'img'
workbook = xlwt.Workbook(encoding='utf-8')
sheet = workbook.add_sheet('票据明细', cell_overwrite_ok=True)

# 2. 写入表头（循环外仅执行一次）
headers = ('年份','出票日期','金额','出票人','付款行全称','汇票到期日','备注')
for col_idx, h in enumerate(headers):
    sheet.write(0, col_idx, h)

count = 1
# 3. 遍历处理所有图片
for root, _, filenames in os.walk(file_dir):
    for fname in filenames:
        img_path = os.path.join(root, fname)
        with open(img_path, 'rb') as f:
            pil_img = PI.open(io.BytesIO(f.read()))
        
        try:
            dt = extract_date(pil_img, tool_tess)
            amt = extract_amount(pil_img, ocr_cn)
            # 其余字段调用对应函数...
            
            sheet.write(count, 0, dt[:4])          # 年份
            sheet.write(count, 1, dt[5:])          # 月日
            sheet.write(count, 2, amt)             # 数值化金额
            # sheet.write(count, 3, txt_drawer)
            # sheet.write(count, 4, txt_payer)
            # ...其他列
            count += 1
        except Exception as e:
            print(f"跳过异常文件 {fname}: {e}")

# 4. 统一保存（放在最终出口）
workbook.save('票据数据汇总.xls')