OCR/config.yaml

# OCR配置文件
output:
  all_results: "./ocr_results/all_results.json"  # 所有识别结果
  current_results: "./ocr_results/current_results.json"  # 当前批次识别结果
  processed_files: "./ocr_results/processed_files.txt"  # 已处理文件列表

# 后端接口配置
upload_url: https://pacs.gw12320.com/adminecg/admin-api/tblist/ecganalysisparas/parsePhotoCreateData
#upload_url: http://localhost:48080/admin-api/tblist/ecganalysisparas/parsePhotoCreateData

# 底部识别配置
bottom_recognition:
  enable: true  # 是否启用底部识别
  height_percent: 20  # 底部区域高度百分比
  width_percent: 100  # 底部区域宽度百分比
  key_words:  # 需要识别的关键字
    - "检查日期"
    - "检查时间"
    - "时间"
    - "日期"
    - "报告日期"
    - "报告时间"
    - "记录时间"
    - "诊断日期"
    - "Date"
    - "Time"
    - "测试日期"
  key_mapping:  # 关键字映射到字段
    "检查日期": "collectDate"
    "检查时间": "collectTime"
    "时间": "collectTime"
    "日期": "collectDate"
    "报告日期": "reportDate"
    "报告时间": "reportTime"
    "记录时间": "recordTime"
    "诊断日期": "diagnosisDate"
    "Date": "date"
    "Time": "time"
    "测试日期": "testDate"

# 图片目录配置
directories:
  - path: "./test_images"
    recognition_type: "normal"
    key_mapping:
      "编 号": "number"
      "姓 名": "name"
      "性 别": "gender"
      "年 龄": "age"
      "科 室": "department"
      "床 号": "bed_number"
      "HR": "HR"
      "PR": "PR"
      "QRS": "QRS"
      "QT/QTC": "QT/QTC"
      "P/QRS/T": "P/QRS/T"
      "RV5/SV1": "RV5/SV1"
      "RV5+SV1": "RV5+SV1"
    bottom_key_words:  # 特定目录的底部关键字配置
      - "检查日期"
      - "日期"
    bottom_key_mapping:  # 特定目录的底部关键字映射
      "检查日期": "checkDate"
      "日期": "date"

  - path: "./ocr_images"
    recognition_type: "normal"
    key_mapping:
      "ID": "id"
      "申请科室": "department"
      "病床号": "bed_number" 
      "HR": "HR"
      "P": "P"
      "QRS": "QRS"
      "QT/QTc": "QT/QTc"
      "P/QRS/T": "P/QRS/T"
      "RV5/SV1": "RV5/SV1"
    bottom_key_words:
      - "检查日期"
      - "时间"
    bottom_key_mapping:
      "检查日期": "examDate"
      "时间": "examTime"

  - path: "./礼泉县裴寨卫生院"  # 分块识别目录 - 3个分块
    recognition_type: "split"
    recognition_area:  # 识别区域配置
      start_x: 0  # 起始X坐标（百分比）
      start_y: 0  # 起始Y坐标（百分比）
      width: 80  # 宽度（百分比）
      height: 30  # 高度（百分比）
    split_blocks:  # 切割块配置
      - width_percent: 25  # 第一块宽度百分比
        key_mapping:
         "编 号": "examId"
         "姓 名": "name"
         "性 别": "gender"
         "年 龄": "age"
         "科 室": "department"
         "床 号": "bed_number"
      - width_percent: 30  # 第二块宽度百分比
        key_mapping:
          "HR": "hr"
          "PR": "pr"
          "QRS": "qrs"
          "QT/QTC": "qt/qtc"
          "P/QRS/T": "pAxle/qrsAxle/tAxle"
          "RV5/SV1": "rv5/sv1"
          "RV5+SV1": "rv5Sv1"
      - width_percent: 45  # 第三块宽度百分比
        key_mapping:
          "备注": "notes"
          "医生": "doctor"
          "日期": "date"
    bottom_key_words:  # 特定目录的底部关键字配置
      - "检查时间"  
    bottom_key_mapping:  # 特定目录的底部关键字映射
      "检查时间": "collectionTime"

  - path: "./special_images"  # 分块识别目录 - 4个分块
    recognition_type: "templateA"
    recognition_area:
      start_x: 0
      start_y: 0
      width: 100
      height: 100
    split_blocks:
      - width_percent: 25
        key_mapping:
          "患者": "patient"
          "ID": "id"
      - width_percent: 25
        key_mapping:
          "检查项目": "exam_item"
          "检查日期": "exam_date"
      - width_percent: 25
        key_mapping:
          "结果1": "result1"
          "结果2": "result2"
      - width_percent: 25
        key_mapping:
          "医师": "doctor"
          "结论": "conclusion"
  - path: "./建陵卫生院"  # 需要旋转90度的图片目录
    recognition_type: "rotate90"  # 新增旋转类型
    recognition_area:  # 添加旋转后的识别区域
      start_x: 0
      start_y: 0
      width: 100  # 宽度百分比
      height: 25  # 高度百分比
    key_mapping:
      "ID": "examId"
      "姓 名": "name"
      "年 龄": "age"
      "性 别": "gender"
      "时 间": "collectionTime"
  - path: "./礼泉县城关卫生院"
    recognition_type: "CG"
    key_mapping:
      "姓名": "name"
      "性别": "gender"
      "年龄": "age"
      "ID": "examId"
      "HR": "hr"
      "P": "P"
      "PR": "pr"
      "QRS": "qrs"
      "QT/QTC": "qt/qtc"
      "P/QRS/T": "pAxle/qrsAxle/tAxle"
      "RV5/SV1": "rv5/sv1"
      "RV5+SV1": "rv5Sv1"
    bottom_key_words:
      - "检查时间"
    bottom_key_mapping:
      "检查时间": "collectionTime"

  - path: "./史德卫生院"
    recognition_type: "LQXSD"
    key_mapping:
      "姓名": "name"
      "性别": "gender"
      "年龄": "age"
      "ID": "examId"
      "HR": "hr"
      "P": "P"
      "PR": "pr"
      "QRS": "qrs"
      "QT/QTC": "qt/qtc"
      "P/QRS/T": "pAxle/qrsAxle/tAxle"
      "RV5/SV1": "rv5/sv1"
      "RV5+SV1": "rv5Sv1"
    bottom_key_words:
      - "检查时间"
    bottom_key_mapping:
      "检查时间": "collectionTime"

# OCR程序与语言包路径配置

# 新增Tesseract相关配置
# bin_path 可选，仅当你需要指定tesseract.exe路径时填写，否则可省略
# data_path 必须，指定tessdata目录的绝对路径
# language 必须，指定语言包

tesseract:
  bin_path: "D:/Program Files/Tesseract-OCR/tesseract.exe"
  data_path: "./tessdata"
  # data_path: "F:/陕西省咸阳市礼泉县心电图FTP/ecgimage/tessdata"
  language: "chi_sim+eng"