#!/usr/bin/env python3 # -*- coding: utf-8 -*- """ 测试OCR识别速度 """ import os import sys import time from pathlib import Path def test_ocr_speed(image_path: str, iterations: int = 5): """测试OCR识别速度""" print("=" * 70) print("OCR速度测试") print("=" * 70) print() # 检查图片文件是否存在 if not os.path.exists(image_path): print(f"[ERROR] 图片文件不存在: {image_path}") return file_size = os.path.getsize(image_path) / 1024 # KB print(f"测试图片: {image_path}") print(f"文件大小: {file_size:.2f} KB") print(f"测试次数: {iterations} 次") print() # 获取图片信息 try: from PIL import Image img = Image.open(image_path) width, height = img.size print(f"图片尺寸: {width}x{height}") print(f"图片模式: {img.mode}") # 计算图片像素数 total_pixels = width * height print(f"总像素数: {total_pixels:,}") # 判断图片大小 if total_pixels > 2000000: # 大于200万像素 print(f" [提示] 图片较大,OCR处理可能需要较长时间") print(f" 建议: 先裁剪到目标区域(如200x200)再OCR,可以显著提升速度") print() except Exception as e: print(f"[WARNING] 无法读取图片信息: {e}") print() # 检查OCR引擎 print("1. OCR引擎信息") print("-" * 70) try: from operations.ocr_operations import _ocr_processor _ocr_processor._ensure_ocr() if _ocr_processor.ocr is None: print("[ERROR] OCR引擎未初始化") return print(f" OCR引擎: {_ocr_processor.ocr_type}") print(f" OMP线程数: {os.environ.get('OMP_NUM_THREADS', '未设置')}") print() except Exception as e: print(f"[ERROR] OCR引擎检查失败: {e}") return # 测试OCR速度 print("2. OCR速度测试") print("-" * 70) times = [] results = [] for i in range(iterations): try: # 导入OCR操作 from Underlying_Operations import underlying_operations ops = underlying_operations() # 确保OCR已初始化 ops._ensure_ocr() # 执行OCR(使用analyse_pic_digits,因为这是数字识别) start_time = time.time() # 直接使用图片路径进行OCR # 注意:如果图片很大,OCR会处理整个图片,这可能会很慢 # 建议使用预处理后的小图片进行测试 result = ops.analyse_pic_digits( "测试图片", change_color=1, # 数字识别模式 pic_path=image_path ) # 记录各个阶段的耗时(如果可能) # 注意:analyse_pic_digits 内部会进行图像预处理和OCR识别 elapsed = (time.time() - start_time) * 1000 # 转换为毫秒 times.append(elapsed) results.append(result) print(f" 第 {i+1} 次: {elapsed:.0f}ms, 识别结果: {result[:50] if result else '(空)'}") except Exception as e: print(f" 第 {i+1} 次: 失败 - {e}") import traceback traceback.print_exc() print() # 统计结果 if times: print("3. 性能统计") print("-" * 70) avg_time = sum(times) / len(times) min_time = min(times) max_time = max(times) print(f" 平均耗时: {avg_time:.0f}ms") print(f" 最快耗时: {min_time:.0f}ms") print(f" 最慢耗时: {max_time:.0f}ms") print(f" 耗时范围: {max_time - min_time:.0f}ms") print() # 识别结果统计 print("4. 识别结果") print("-" * 70) unique_results = list(set(results)) for i, result in enumerate(unique_results, 1): count = results.count(result) print(f" 结果 {i}: {result[:100] if result else '(空)'} (出现 {count} 次)") print() # 性能评估 print("5. 性能评估") print("-" * 70) if avg_time < 500: print(" [优秀] OCR速度很快 (< 500ms)") elif avg_time < 1000: print(" [良好] OCR速度良好 (500-1000ms)") elif avg_time < 2000: print(" [一般] OCR速度一般 (1000-2000ms)") elif avg_time < 5000: print(" [较慢] OCR速度较慢 (2000-5000ms)") else: print(" [很慢] OCR速度很慢 (> 5000ms)") print(" 提示: 如果图片很大,建议先裁剪到目标区域再OCR") # 多线程效果评估 cpu_count = os.cpu_count() or 4 omp_threads = int(os.environ.get('OMP_NUM_THREADS', '1')) print(f" CPU核心数: {cpu_count}") print(f" OMP线程数: {omp_threads}") if omp_threads > 1: print(f" [优化] 已启用底层多线程 (OMP_NUM_THREADS={omp_threads})") else: print(f" [提示] 未启用底层多线程,可以优化") print() else: print("[ERROR] 所有测试都失败了") print("=" * 70) print("测试完成") print("=" * 70) def main(): """主函数""" # 默认图片路径 default_image = "pic/debug_captures/2025-12-31/20251231_152116_ocr_福袋倒计时_秒_预处理后_(480,1110)-(570,1230)_模式_1_尺寸_270x360_来源_福袋倒计时_秒.png_有完整截图_是_有坐标_是_类型_digits.png" # 从命令行参数获取图片路径 if len(sys.argv) > 1: image_path = sys.argv[1] else: image_path = default_image # 获取测试次数 iterations = 5 if len(sys.argv) > 2: try: iterations = int(sys.argv[2]) except: pass # 转换为绝对路径 if not os.path.isabs(image_path): # 尝试相对于项目根目录 project_root = os.path.dirname(os.path.abspath(__file__)) image_path = os.path.join(project_root, image_path) test_ocr_speed(image_path, iterations) if __name__ == "__main__": main()