删除pdf扫描件里的空白页

使用场景：

报告里有两面，单面的，想扫描成pdf，统一双面扫描，然后删除空白页

import os import fitz import numpy as np base_dir = r"E:\公司文件\报告资料归档整理\开平农村公路55座桥梁检测服务BGQL202500146~158" output_dir = os.path.join(base_dir, "已删除空白页的文件") os.makedirs(output_dir, exist_ok=True) def is_center_mostly_white(page, center_ratio=0.7, white_threshold=0.98): """ 仅检查页面中心区域是否几乎全白（忽略页眉页脚/边框） """ mat = fitz.Matrix(1.0, 1.0) # 速度优先 pix = page.get_pixmap(matrix=mat, alpha=False) img = np.frombuffer(pix.samples, dtype=np.uint8).reshape(pix.height, pix.width, 3) h, w = img.shape[:2] margin_h = int(h * (1 - center_ratio) / 2) margin_w = int(w * (1 - center_ratio) / 2) center_img = img[margin_h:-margin_h, margin_w:-margin_w] is_white = np.all(center_img >= 245, axis=2) # 更严格：≥245 white_ratio = np.mean(is_white) return white_ratio >= white_threshold # 获取所有 BGQL*.pdf 文件（自动扫描） import glob pdf_files = [os.path.join(base_dir, f"BGQL202500{i}.pdf") for i in range(158, 159)] #注意range(146, 158)不包含158 for input_path in pdf_files: filename = os.path.basename(input_path) try: doc = fitz.open(input_path) new_doc = fitz.open() removed = 0 for page_num in range(len(doc)): page = doc[page_num] # 跳过有文字的页（防误删） if page.get_text("text").strip(): new_doc.insert_pdf(doc, from_page=page_num, to_page=page_num) continue # 无文字时，检查中心是否全白 if is_center_mostly_white(page, center_ratio=0.7, white_threshold=0.98): print(f" → {filename} 第 {page_num + 1} 页中心全白，删除") removed += 1 else: new_doc.insert_pdf(doc, from_page=page_num, to_page=page_num) output_path = os.path.join(output_dir, filename) new_doc.save(output_path, garbage=4, deflate=True) new_doc.close() doc.close() print(f"✅ {filename}: 删除 {removed} 页") except Exception as e: print(f"❌ 处理 {filename} 出错: {e}") print(f"\n🔚 视觉空白清理完成！结果: {output_dir}")