|
版本更新第五版, 版本号1.0.5
1.0.4更新内容: 模型改为随程序分发, 初次启动时无需再下载模型, 解决黑框终端无法关闭的问题
1.0.5更新内容: 优化字段匹配方法, 识别精确度有所提高
版本更新第三版, 代码已开源(见附件)
更新内容: 优化软件启动速度, 增加识别字段, 增加图片重命名(可选可配置), 增加excel内图片选项(路径/图片)
由于代码量较大, 无法在此处展开, 有对源码或进一步优化感兴趣的伙伴可以下载源码阅读修改
下面的代码我只放核心逻辑部分
由于一些行业需要手动录入大量的身份证信息, 因此编写本软件用于减少工作
软件说明: 本软件使用python3.12.10编写, 因此无法在win10以下的电脑上运行，gui改为pySide6，识别功能依赖包：paddlepaddle==2.6.2，paddleocr，paddlenlp，调用cpu进行识别，不依赖显卡。如果你的电脑显卡比较好，可以将依赖改为GPU版，再进行打包，使用gpu性能会更强。
代码提供Windows版本，由于ocr模型在本地运行，因此软件对电脑性能要求较高，我Linux云服务器则是2核2G，带不动。Windows电脑一般不会有那么低的配置，则不用担心。可断网进行识别。
代码如下：
[Python] 纯文本查看 复制代码# 在gui.py文件中添加以下代码import re
from PySide6.QtCore import QThread, Signalimport tracebackfrom openpyxl.workbook import Workbookfrom openpyxl.drawing.image import Image as XLImagefrom openpyxl.styles import Alignment
import cv2import numpy as npfrom PIL import Image
'''ocr实际开始工作的线程需要将前边加载好的模型传递过来'''class OCRWorker(QThread):
定义信号，用于通知主线程处理进度和结果
progress_updated = Signal(int, int) # 当前进度，总数量finished_signal = Signal() # 处理完成信号error_occurred = Signal(str) # 错误信息信号def __init__(self, file_paths, export_options, ocr): super().__init__() self.file_paths = file_paths self.export_options = export_options self.ocr = ocr self._should_terminate = False # 添加终止标志def run(self): try: # 处理所有文件 self.process_files(self.file_paths) except Exception as e: error_msg = f"处理过程中发生错误: {str(e)}\n{traceback.format_exc()}" self.error_occurred.emit(error_msg)def process_files(self, file_paths): wb = Workbook() ws = wb.active ws.append(["图片", "姓名", "性别", "民族", "出生日期", "住址", "身份证号", "有效期限"]) row_idx = 2 total_files = len(self.file_paths) processed_count = 0 for i, path in enumerate(file_paths): # 检查是否收到终止请求 if self._should_terminate: print("收到终止请求，正在保存已处理的数据...") break # 发送进度更新信号 self.progress_updated.emit(i + 1, total_files) info = self.extract_info_from_image(path) # 检查线程是否被中断 if self.isInterruptionRequested(): break if info: ws.cell(row=row_idx, column=2, value=info["姓名"]) ws.cell(row=row_idx, column=3, value=info["性别"]) ws.cell(row=row_idx, column=4, value=info["民族"]) ws.cell(row=row_idx, column=5, value=info["出生日期"]) ws.cell(row=row_idx, column=6, value=info["住址"]) ws.cell(row=row_idx, column=7, value=info["身份证号"]) ws.cell(row=row_idx, column=8, value=info["有效期限"]) # 根据导出选项决定如何处理图片 export_option = self.export_options.get("export_option", "image_path") # 处理重命名（如果配置了重命名选项） if self.should_rename_file(info): new_path = self.rename_file(path, info) # 更新图片路径为重命名后的路径 if export_option == "image_path": ws.cell(row=row_idx, column=1, value=new_path) # 更新返回数据中的图片路径 info["图片路径"] = new_path if export_option == "image_file": # 直接嵌入图片文件 try: img = XLImage(info["图片路径"]) img.width = 500 img.height = 300 ws.row_dimensions[zxsq-anti-bbcode-row_idx].height = img.height ws.add_image(img, f"A{row_idx}") ws.column_dimensions['A'].width = img.width * 0.14 except Exception as e: print(f"无法插入图片 {path}: {e}") else : # 仅保存图片路径 ws.cell(row=row_idx, column=1, value=info["图片路径"]) for col in range(1, 9): cell = ws.cell(row=row_idx, column=col) cell.alignment = Alignment(horizontal='center', vertical='center') row_idx += 1 processed_count += 1 for col in range(1, 9): header_cell = ws.cell(row=1, column=col) header_cell.alignment = Alignment(horizontal='center', vertical='center') output_path = "身份证识别结果.xlsx" wb.save(output_path) if self._should_terminate: print(f"处理已终止，已完成 {processed_count}/{total_files} 个文件，结果已保存到 {output_path}") else: print(f"处理完成，共处理 {processed_count} 个文件，结果已保存到 {output_path}") # 发送完成信号 self.finished_signal.emit()def extract_info_from_image(self, image_path): """从图片中提取信息（优化版文本处理）""" try: # 检查文件是否存在和可读 # import os # if not os.path.exists(image_path): # raise FileNotFoundError(f"图片文件不存在: {image_path}") # # if not os.access(image_path, os.R_OK): # raise PermissionError(f"没有权限读取图片文件: {image_path}") # # 检查是否需要预处理身份证图片 # if self.export_options.get("preprocess_id_card", True): # processed_image_path = self.preprocess_id_card_image(image_path) # else: # processed_image_path = image_path # # result = self.ocr.ocr(processed_image_path, cls=True) result = self.ocr.ocr(image_path, cls=True) # 1. 先整体拼接所有文本 all_text = "" for res in result: for line in res: text = line[zxsq-anti-bbcode-1][zxsq-anti-bbcode-0] if text: all_text += text # 2. 去除"中华人民共和国居民身份证"标题 all_text = re.sub(r'中华人民共和国居民身份证', '', all_text) # 3. 去除所有空格和特殊空白字符 all_text = re.sub(r'\s+', '', all_text) # 4. 在关键字段前添加换行符 keywords = ['姓名', '性别', '民族', '出生', '住址', '公民身份号码', '签发机关', '有效期限'] for keyword in keywords: all_text = re.sub(f'({keyword})', r'\n\1', all_text) print(f"处理后的文本: {all_text}") # 初始化提取结果 name = gender = nation = birth = address = id_number = expire = "" # 提取各字段信息 # 提取身份证号 # 直接匹配17位数字+1位校验码（数字或X） id_match = re.search(r'[\d]{17}[\dXx]', all_text) if id_match: id_number = id_match.group().strip() # 移除身份证号码干扰 all_text = all_text.replace(id_match.group(), '') # 提取姓名 name_match = re.search(r'姓名(.+?)(?=\n|$)', all_text) if name_match: name = name_match.group(1).strip() # 提取性别 gender_match = re.search(r'性别(男|女)', all_text) if gender_match: gender = gender_match.group(1).strip() # 提取民族 nation_match = re.search(r'民族(.+?)(?=\n|$)', all_text) if nation_match: nation = nation_match.group(1).strip() # 提取出生日期 birth_match = re.search(r'出生(.+?)(?=\n|$)', all_text) if birth_match: birth = birth_match.group(1).strip() # 提取住址 address_match = re.search(r'住址(.+?)(?=\n|$)', all_text) if address_match: address = address_match.group(1).strip() # 提取有效期限 expire_match = re.search(r'有效期限(.+?)(?=\n|$)', all_text) if expire_match: expire = expire_match.group(1).strip() data = { "姓名": name, "性别": gender, "民族": nation, "出生日期": birth, "住址": address, "身份证号": id_number, "有效期限": expire, "图片路径": image_path } print(f"data == {data}") return data except Exception as e: print(f"处理 {image_path} 失败: {e}") return Nonedef should_rename_file(self, info): """检查是否需要重命名文件""" rename_options = self.export_options.get("rename_options", []) return len(rename_options) > 0def rename_file(self, original_path, info): """根据配置重命名文件""" if not self.should_rename_file(info): return original_path rename_options = self.export_options.get("rename_options", []) separator = self.export_options.get("separator", "_") # 构建新的文件名部分 name_parts = [] for option in rename_options: if option == "name" and info.get("姓名"): name_parts.append(info["姓名"]) elif option == "id" and info.get("身份证号"): name_parts.append(info["身份证号"]) elif option == "nation" and info.get("民族"): name_parts.append(info["民族"]) elif option == "sex" and info.get("性别"): name_parts.append(info["性别"]) elif option == "address" and info.get("住址"): name_parts.append(info["住址"]) if not name_parts: return original_path # 构造新文件名 new_name = separator.join(name_parts) # 保持原始文件扩展名 import os dir_name = os.path.dirname(original_path) file_ext = os.path.splitext(original_path)[zxsq-anti-bbcode-1] new_path = os.path.join(dir_name, new_name + file_ext) # 重命名文件 try: os.rename(original_path, new_path) return new_path except Exception as e: print(f"重命名文件失败 {original_path} -> {new_path}: {e}") return original_path# 图片灰度处理, 处理成扫描件, 下面还没写好 不要用# def preprocess_id_card_image(self, image_path):# """对身份证图片进行校正、裁剪并转换为黑白扫描件"""# try:# # 读取图片# img = cv2.imread(image_path)# if img is None:# return image_path## # 1. 转换为灰度图# gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)## # 2. 使用中值滤波代替高斯模糊# denoised = cv2.medianBlur(gray, 3)## # 3. 使用自适应阈值# binary = cv2.adaptiveThreshold(# denoised, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C,# cv2.THRESH_BINARY, 15, 3# )## # 4. 可选：轻微平滑处理# smoothed = cv2.medianBlur(binary, 1)## # 5. 保存处理后的图片# import os# dir_name = os.path.dirname(image_path)# file_name = os.path.splitext(os.path.basename(image_path))[zxsq-anti-bbcode-0]# file_ext = os.path.splitext(image_path)[zxsq-anti-bbcode-1]# processed_path = os.path.join(dir_name, f"{file_name}_processed{file_ext}")## cv2.imwrite(processed_path, smoothed)## return processed_path# except Exception as e:# print(f"身份证图片预处理失败 {image_path}: {e}")# return image_path## def order_points(self, pts):# """对四个点进行排序：左上、右上、右下、左下"""# rect = np.zeros((4, 2), dtype="float32")## # 计算坐标和# s = pts.sum(axis=1)# rect[zxsq-anti-bbcode-0] = pts[np.argmin(s)] # 左上角点（坐标和最小）# rect[zxsq-anti-bbcode-2] = pts[np.argmax(s)] # 右下角点（坐标和最大）## # 计算坐标差# diff = np.diff(pts, axis=1)# rect[zxsq-anti-bbcode-1] = pts[np.argmin(diff)] # 右上角点（坐标差最小）# rect[zxsq-anti-bbcode-3] = pts[np.argmax(diff)] # 左下角点（坐标差最大）## return rect## def four_point_transform(self, image, pts):# """四点透视变换"""# # 获取排序后的坐标# rect = self.order_points(pts)# (tl, tr, br, bl) = rect## # 计算新图像的宽度和高度# width_a = np.sqrt(((br[zxsq-anti-bbcode-0] - bl[zxsq-anti-bbcode-0]) ** 2) + ((br[zxsq-anti-bbcode-1] - bl[zxsq-anti-bbcode-1]) ** 2))# width_b = np.sqrt(((tr[zxsq-anti-bbcode-0] - tl[zxsq-anti-bbcode-0]) ** 2) + ((tr[zxsq-anti-bbcode-1] - tl[zxsq-anti-bbcode-1]) ** 2))# max_width = max(int(width_a), int(width_b))## height_a = np.sqrt(((tr[zxsq-anti-bbcode-0] - br[zxsq-anti-bbcode-0]) ** 2) + ((tr[zxsq-anti-bbcode-1] - br[zxsq-anti-bbcode-1]) ** 2))# height_b = np.sqrt(((tl[zxsq-anti-bbcode-0] - bl[zxsq-anti-bbcode-0]) ** 2) + ((tl[zxsq-anti-bbcode-1] - bl[zxsq-anti-bbcode-1]) ** 2))# max_height = max(int(height_a), int(height_b))## # 目标点# dst = np.array([# [0, 0],# [max_width - 1, 0],# [max_width - 1, max_height - 1],# [0, max_height - 1]], dtype="float32")## # 计算透视变换矩阵并应用# M = cv2.getPerspectiveTransform(rect, dst)# warped = cv2.warpPerspective(image, M, (max_width, max_height))## return warped# 中断处理, 此处不要直接中断线程, 可能导致excel未能处理完毕线程就退出了# 我们应该保证exceldef request_termination(self): """请求终止处理过程""" self._should_terminate = True
下载链接.txt
(27.9 KB, 下载次数: 0)
|
重要声明：以上内容仅代表作者戴明利观点，不代表本站测量协会立场。如有涉及侵权请尽快告知，我们将会在第一时间处理。作者原创内容未经允许不得转载！
站长联系邮箱：1339305021@qq.com
站长联系微信：dddnnbbb
|
桂ICP备2026007449号-1 )
窥视卡
雷达卡
发表于 2025-5-22 10:10:00
提升卡
置顶卡
沉默卡
喧嚣卡
变色卡
千斤顶
照妖镜