Flow/OCR
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge branch 'main' of http://123.57.244.246:3000/Flow/OCR

Browse Source
This commit is contained in:
Flow 2025-05-15 10:01:14 +08:00
commit e277547c6e
2 changed files with 475 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

46
config.yaml
View File

@ -5,8 +5,8 @@ output:
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
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:
@ -141,7 +141,6 @@ directories:
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
@ -155,6 +154,45 @@ directories:
"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
@ -164,7 +202,7 @@ directories:
# language 必须，指定语言包
tesseract:
bin_path: "C:/Program Files/Tesseract-OCR/tesseract.exe"
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"

433
src/main/java/com/ocr/FolderMonitor.java
View File

@ -89,6 +89,12 @@ public class FolderMonitor {
case "templateA":
extractedData = processImageWithTemplateA(imageFullPath);
break;
case "CG":
extractedData = processImageWithCG(imageFullPath);
break;
case "LQXSD":
extractedData = processImageWithLQXSD(imageFullPath);
break;
case "rotate90":
extractedData = processImageWithRotate90(imageFullPath);
break;
@ -705,6 +711,433 @@ public class FolderMonitor {
return processImageNormal(imageFullPath);
}
// 新版CG识别逻辑：按区域裁剪后分别OCR提取字段
private Map<String, String> processImageWithCG(String imageFullPath) {
Map<String, String> extractedData = new HashMap<>();
try {
BufferedImage image = ImageIO.read(new File(imageFullPath));
int width = image.getWidth();
int height = image.getHeight();
// 1. 标题区和患者信息区分开裁剪
int titleHeight = (int) (height * 0.01); // 标题区5%
int infoHeight = (int) (height * 0.10); // 患者信息区10%
// 跳过标题区，只识别患者信息区
BufferedImage infoArea = image.getSubimage(0, titleHeight, width, infoHeight);
String infoOcr = tesseract.doOCR(infoArea);
String[] infoLines = infoOcr.split("\\r?\\n");
String namePattern = "^([\\u4e00-\\u9fa5 ]+)\\s+(女|男)";
String agePattern = "(\\d+)\\s*[岁%]";
String idPattern = "[I1l][D][:：]?\\s*([A-Za-z0-9]+)";
for (String line : infoLines) {
line = line.replaceAll("\\s+", " ").trim();
if (extractedData.get("name") == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(namePattern).matcher(line);
if (m.find()) {
extractedData.put("name", m.group(1).replaceAll(" ", "")); // 去除姓名中的空格
extractedData.put("gender", m.group(2));
java.util.regex.Matcher ageM = java.util.regex.Pattern.compile(agePattern).matcher(line);
if (ageM.find()) {
extractedData.put("age", ageM.group(1));
}
}
}
if (extractedData.get("id") == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(idPattern).matcher(line);
if (m.find()) {
extractedData.put("id", m.group(1));
}
}
}
// 2. 左侧参数区（提取HR、P、PR、QRS、QT/QTC、P/QRS/T、RV5/SV1）
int paramWidth = (int) (width * 0.32); // 左侧32%
int paramStartY = titleHeight;
int paramHeight = (int) (height * 0.355); // 参数区高度约38%
BufferedImage paramArea = image.getSubimage(0, paramStartY, paramWidth, paramHeight);
String paramOcr = tesseract.doOCR(paramArea);
String[] paramLines = paramOcr.split("\\r?\\n");
String hrPattern = "HR\\s*[:：.·]?\\s*([\\dOo./]+[bB][pP][mM])";
String pPattern = "P\\s*[:：.·]?\\s*([\\dOo./]+[mM][sS])";
String prPattern = "PR\\s*[:：.·]?\\s*([\\dOo./]+[mM][sS])";
String qrsPattern = "QRS\\s*[:：.·]?\\s*([\\dOo./]+[mM][sS])";
String qtPattern = "(?:QT|QI)/QT[cC]?\\s*[:：.·]?\\s*([\\dOo./]+[mM][sS])";
String qtcPattern = "[@Q]Tc\\s*间\\s*期\\s*[:：]?\\s*(\\d+)\\s*[mM][sS]";
String pqrstPattern = "P/QRS/T\\s*[:：.·]?\\s*([\\dOo./]+)";
String rv5sv1Pattern = "([_\\s]*[Rr][Vv5][Ss][\\s/]*[Ss][Vv1Ii])\\s*[:：.·]?\\s*([\\dOo./]+)\\s*[mM][vVyY]?";
for (String line : paramLines) {
line = line.replaceAll("\\s+", " ").trim();
if (extractedData.get("hr") == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(hrPattern, java.util.regex.Pattern.CASE_INSENSITIVE).matcher(line);
if (m.find()) {
String hrValue = m.group(1).replaceAll("[Oo]", "0").replaceAll("[^\\d.]", "");
extractedData.put("hr", hrValue);
}
}
if (extractedData.get("p") == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(pPattern, java.util.regex.Pattern.CASE_INSENSITIVE).matcher(line);
if (m.find()) {
String pValue = m.group(1).replaceAll("[Oo]", "0").replaceAll("[^\\d.]", "");
extractedData.put("p", pValue);
}
}
if (extractedData.get("pr") == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(prPattern, java.util.regex.Pattern.CASE_INSENSITIVE).matcher(line);
if (m.find()) {
String prValue = m.group(1).replaceAll("[Oo]", "0").replaceAll("[^\\d.]", "");
extractedData.put("pr", prValue);
}
}
if (extractedData.get("qrs") == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(qrsPattern, java.util.regex.Pattern.CASE_INSENSITIVE).matcher(line);
if (m.find()) {
String qrsValue = m.group(1).replaceAll("[Oo]", "0").replaceAll("[^\\d.]", "");
extractedData.put("qrs", qrsValue);
}
}
if (extractedData.get("qt/qtc") == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(qtPattern, java.util.regex.Pattern.CASE_INSENSITIVE).matcher(line);
if (m.find()) {
String qtValue = m.group(1).replaceAll("[Oo]", "0").replaceAll("[^\\d./]", "");
extractedData.put("qt/qtc", qtValue);
}
}
if (extractedData.get("pAxle/qrsAxle/tAxle") == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(pqrstPattern, java.util.regex.Pattern.CASE_INSENSITIVE).matcher(line);
if (m.find()) {
String pqrstValue = m.group(1).replaceAll("[Oo]", "0").replaceAll("[^\\d/degDEG.]", "");
extractedData.put("pAxle/qrsAxle/tAxle", pqrstValue);
}
}
if (extractedData.get("rv5/sv1") == null) {
if (line.toLowerCase().contains("sv1")) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile("([\\dOo.]+)/([\\dOo.]+)\\s*[mM][vVyY]").matcher(line);
if (m.find()) {
String rv5 = m.group(1).replaceAll("[Oo]", "0");
String sv1 = m.group(2).replaceAll("[Oo]", "0");
String rv5sv1Value = rv5 + "/" + sv1;
extractedData.put("rv5/sv1", rv5sv1Value);
// 计算rv5Sv1的值（rv5和sv1的和）
try {
double rv5Value = Double.parseDouble(rv5);
double sv1Value = Double.parseDouble(sv1);
double sum = rv5Value + sv1Value;
extractedData.put("rv5Sv1", String.format("%.3f", sum));
} catch (NumberFormatException e) {
logger.error("计算rv5Sv1时发生数字格式错误: rv5={}, sv1={}", rv5, sv1, e);
}
}
}
}
}
// 3. 底部区域（提取检查时间）
int bottomHeight = (int) (height * 0.05); // 底部12%
int bottomStartY = height - bottomHeight;
BufferedImage bottomArea = image.getSubimage(0, bottomStartY, width, bottomHeight);
String bottomOcr = tesseract.doOCR(bottomArea);
String[] bottomLines = bottomOcr.split("\\r?\\n");
String checkTimePattern = "检查[:：]?\\s*([\\d-]+ [\\d:]+)";
for (String line : bottomLines) {
String lineNoSpace = line.replaceAll("\\s+", "");
if (extractedData.get("collectionTime") == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile("检查[:：]?(\\d{4}-\\d{2}-\\d{2}\\s*\\d{2}:\\d{2}:\\d{2})").matcher(lineNoSpace);
if (m.find()) {
String dateTime = m.group(1);
if (!dateTime.contains(" ")) {
dateTime = dateTime.substring(0, 10) + " " + dateTime.substring(10);
}
extractedData.put("collectionTime", dateTime);
}
}
}
} catch (Exception e) {
logger.error("processImageWithCG 区域识别异常", e);
}
return extractedData;
}
// 新增礼泉县夹德中心卫生院模板识别方法
private Map<String, String> processImageWithLQXSD(String imageFullPath) {
Map<String, String> extractedData = new HashMap<>();
try {
BufferedImage image = ImageIO.read(new File(imageFullPath));
int width = image.getWidth();
int height = image.getHeight();
// 1. 顶部患者信息区（约前12%）
int infoHeight = (int) (height * 0.12);
BufferedImage infoArea = image.getSubimage(0, 0, width, infoHeight);
String infoOcr = tesseract.doOCR(infoArea);
String[] infoLines = infoOcr.split("\\r?\\n");
String namePattern = "[姓姊]\\s*名\\s*[:：]?\\s*([\\u4e00-\\u9fa5]+)(?=\\s*[门]?诊\\s*号|\\s*性\\s*别|$)";
String genderPattern = "性\\s*别\\s*[:：]?\\s*([男女])";
String agePattern = "年\\s*龄\\s*[:：]?\\s*(\\d+)\\s*[岁%]";
String idPattern = "序\\s*号\\s*[:；;]?\\s*([A-Za-z0-9]+)";
for (String line : infoLines) {
line = line.trim();
// 姓名
if (extractedData.get("name") == null) {
// 先尝试使用正则表达式匹配
java.util.regex.Matcher m = java.util.regex.Pattern.compile(namePattern).matcher(line);
if (m.find()) {
String name = m.group(1).replaceAll("\\s+", "");
if (name.contains("门诊号")) {
name = name.split("门诊号")[0];
}
extractedData.put("name", name);
} else {
// 如果正则匹配失败，尝试直接提取
if (line.contains("姓名") || line.contains("姓名：") || line.contains("姓 名") ||
line.contains("姊名") || line.contains("姊名：") || line.contains("姊 名")) {
// 统一处理各种可能的姓名标识
String processedLine = line.replaceAll("[姓姊]\\s*名\\s*[:：]?", "姓名:");
String[] parts = processedLine.split("姓名:");
if (parts.length > 1) {
String afterName = parts[1].trim();
// 找到下一个字段的位置
int nextFieldIndex = -1;
if (afterName.contains("门诊号") || afterName.contains("门 诊 号")) {
nextFieldIndex = afterName.indexOf("门诊号");
if (nextFieldIndex == -1) {
nextFieldIndex = afterName.indexOf("门 诊 号");
}
} else if (afterName.contains("性别") || afterName.contains("性 别")) {
nextFieldIndex = afterName.indexOf("性别");
if (nextFieldIndex == -1) {
nextFieldIndex = afterName.indexOf("性 别");
}
}
if (nextFieldIndex > 0) {
String name = afterName.substring(0, nextFieldIndex)
.replaceAll("[:：]", "")
.replaceAll("\\s+", "")
.trim();
extractedData.put("name", name);
}
}
}
}
}
// 性别
if (extractedData.get("gender") == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(genderPattern).matcher(line);
if (m.find()) {
extractedData.put("gender", m.group(1));
}
}
// 年龄
if (extractedData.get("age") == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(agePattern).matcher(line);
if (m.find()) {
extractedData.put("age", m.group(1));
}
}
// ID
if (extractedData.get("examId") == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(idPattern).matcher(line);
if (m.find()) {
extractedData.put("examId", m.group(1));
}
}
// 识别日期
if (extractedData.get("collectionTime") == null) {
// 处理原有格式和带空格的格式
String datePattern = "日\\s*期\\s*[:：]?\\s*(\\d{4}[-/]\\d{1,2}[-/]\\d{1,2})";
// 预处理行，处理字符间可能有空格的情况
String processedLine = line.replaceAll("\\s+", "");
// 先尝试原始行
java.util.regex.Matcher m = java.util.regex.Pattern.compile(datePattern).matcher(line);
if (m.find()) {
String dateValue = m.group(1).replaceAll("/", "-");
extractedData.put("collectionTime", dateValue);
} else {
// 如果原始行匹配失败，尝试处理后的行
m = java.util.regex.Pattern.compile(datePattern).matcher(processedLine);
if (m.find()) {
String dateValue = m.group(1).replaceAll("/", "-");
extractedData.put("collectionTime", dateValue);
}
}
}
}
// 2. 参数区（底部1/3，适配新模板）
int paramStartY = (int) (height * 0.81);
int paramHeight = height - paramStartY;
BufferedImage paramArea = image.getSubimage(0, paramStartY, width, paramHeight);
String paramOcr = tesseract.doOCR(paramArea);
String[] paramLines = paramOcr.split("\\r?\\n");
String hrPattern = "心率[:：]?\\s*([\\dOo./]+)";
String pPattern = "P\\s*(?:时限)?[:：]?\\s*([\\dOo./+-]+)(?:\\s*[mM][sS])?";
String pAxlePattern = "P\\s*(?:轴)?[:：]?\\s*([\\dOo./+-]+)\\s*°";
String prPattern = "PR间期[:：;]?\\s*([\\dOo./]+)";
String qrsPattern = "QRS时限[:：]?\\s*([\\dOo./]+)";
String qtPattern = "(?:[@Q]?T|QT)\\s*间\\s*期\\s*[:：.·]?\\s*(\\d+)\\s*[mM][sS]";
String qtcPattern = "(?:[@Q]Tc|afe)\\s*间\\s*期\\s*[:：.·]?\\s*(\\d+)\\s*[mM][sS]";
String qrsAxlePattern = "QRS轴[:：]?\\s*([\\dOo.]+)";
String tAxlePattern = "T轴[:：]?\\s*([\\dOo.]+)";
String rv5sv1Pattern = "RV5/SV1.*?([\\dOo.]+/[\\dOo.]+|[\\dOo.]+).*?[mM][vVyY]?";
String rv5plusSv1Pattern = "RV5\\+SV1[:：]?\\s*([\\dOo./]+)";
// 用于存储QT和QTc的值
String qtValue = null;
String qtcValue = null;
// 用于存储三个电轴的值
String pAxleValue = null;
String qrsAxleValue = null;
String tAxleValue = null;
// 标记是否已识别为P轴
boolean isPAxleDetected = false;
for (String line : paramLines) {
line = line.replaceAll("\\s+", "").trim();
// HR
if (extractedData.get("hr") == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(hrPattern).matcher(line);
if (m.find()) {
String hrValue = m.group(1).replaceAll("[Oo]", "0");
extractedData.put("hr", hrValue);
}
}
// 处理所有P值 - 先检查是否含有度数符号判断是P轴还是P时限
if (line.contains("P") && line.contains("°")) {
// 如果行中同时包含P和度数符号，尝试提取P轴值
java.util.regex.Matcher m = java.util.regex.Pattern.compile(pAxlePattern).matcher(line);
if (m.find() && pAxleValue == null) {
pAxleValue = m.group(1).replaceAll("[Oo]", "0");
isPAxleDetected = true; // 标记已识别为P轴
}
} else if ((line.contains("P") && !line.contains("PR") && !isPAxleDetected) || line.contains("P时限")) {
// 不含度数符号且不是PR，或明确包含"P时限"字样，尝试提取P时限
java.util.regex.Matcher m = java.util.regex.Pattern.compile(pPattern).matcher(line);
if (m.find() && extractedData.get("P") == null) {
String pValue = m.group(1).replaceAll("[Oo]", "0");
extractedData.put("P", pValue);
}
}
// P - 仅当未识别为P轴时才尝试识别为P时限，但明确包含"P时限"的行除外
if ((extractedData.get("P") == null && !isPAxleDetected) || line.contains("P时限")) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(pPattern).matcher(line);
if (m.find()) {
String pValue = m.group(1).replaceAll("[Oo]", "0");
extractedData.put("P", pValue);
}
}
// P轴 - 检查传统格式，仅当P轴未识别时
if (pAxleValue == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile("P轴[:：]?\\s*([\\dOo.+-]+)").matcher(line);
if (m.find()) {
pAxleValue = m.group(1).replaceAll("[Oo]", "0");
isPAxleDetected = true; // 标记已识别为P轴
}
}
// PR
if (extractedData.get("pr") == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(prPattern).matcher(line);
if (m.find()) {
extractedData.put("pr", m.group(1).replaceAll("[Oo]", "0"));
}
}
// QRS
if (extractedData.get("qrs") == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(qrsPattern).matcher(line);
if (m.find()) {
extractedData.put("qrs", m.group(1).replaceAll("[Oo]", "0"));
}
}
// QT
if (qtValue == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(qtPattern).matcher(line);
if (m.find()) {
qtValue = m.group(1).replaceAll("[Oo]", "0");
}
}
// QTc
if (qtcValue == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(qtcPattern).matcher(line);
if (m.find()) {
qtcValue = m.group(1).replaceAll("[Oo]", "0");
}
}
// QRS轴
if (qrsAxleValue == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(qrsAxlePattern).matcher(line);
if (m.find()) {
qrsAxleValue = m.group(1).replaceAll("[Oo]", "0");
}
}
// T轴
if (tAxleValue == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(tAxlePattern).matcher(line);
if (m.find()) {
tAxleValue = m.group(1).replaceAll("[Oo]", "0");
}
}
// RV5/SV1
if (extractedData.get("rv5/sv1") == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(rv5sv1Pattern).matcher(line);
if (m.find()) {
extractedData.put("rv5/sv1", m.group(1).replaceAll("[Oo]", "0"));
}
}
// RV5+SV1
if (extractedData.get("rv5Sv1") == null) {
java.util.regex.Matcher m = java.util.regex.Pattern.compile(rv5plusSv1Pattern).matcher(line);
if (m.find()) {
extractedData.put("rv5Sv1", m.group(1).replaceAll("[Oo]", "0"));
}
}
}
// 处理QT/QTc的值
if (qtValue != null || qtcValue != null) {
if (qtValue != null && qtcValue != null) {
extractedData.put("qt/qtc", qtValue + "/" + qtcValue);
} else if (qtValue != null) {
extractedData.put("qt/qtc", qtValue);
} else if (qtcValue != null) {
extractedData.put("qt/qtc", qtcValue);
}
}
// 处理三个电轴的值
if (pAxleValue != null || qrsAxleValue != null || tAxleValue != null) {
StringBuilder axleValue = new StringBuilder();
if (pAxleValue != null) {
axleValue.append(pAxleValue);
}
if (qrsAxleValue != null) {
if (axleValue.length() > 0) {
axleValue.append("/");
}
axleValue.append(qrsAxleValue);
}
if (tAxleValue != null) {
if (axleValue.length() > 0) {
axleValue.append("/");
}
axleValue.append(tAxleValue);
}
extractedData.put("pAxle/qrsAxle/tAxle", axleValue.toString());
}
} catch (Exception e) {
logger.error("processImageWithLQXSD 识别异常", e);
}
return extractedData;
}
/**
* 处理图片，将图片逆时针旋转90度
* @param imageFullPath 图片完整路径