""" OCR 模块使用示例 演示如何使用 OCR 模块进行文字识别 """ # 导入 OCR 模块 from src.core.ocr import ( recognize_text, preprocess_image, PaddleOCREngine, CloudOCREngine, OCRFactory, ImagePreprocessor, OCRResult, OCRLanguage ) def example_1_quick_recognize(): """示例 1: 快速识别文本（最简单）""" print("示例 1: 快速识别文本") print("-" * 50) result = recognize_text( image="path/to/your/image.png", mode="local", # 本地识别 lang="ch", # 中文 use_gpu=False, # 不使用 GPU preprocess=False # 不预处理 ) if result.success: print(f"识别成功！") print(f"平均置信度: {result.total_confidence:.2f}") print(f"识别行数: {len(result.results)}") print(f"完整文本:\n{result.full_text}") else: print(f"识别失败: {result.error_message}") def example_2_with_preprocess(): """示例 2: 带预处理的识别（适合低质量图片）""" print("\n示例 2: 带预处理的识别") print("-" * 50) result = recognize_text( image="path/to/your/image.png", mode="local", lang="ch", preprocess=True # 启用预处理（增强对比度、锐度等） ) if result.success: print(f"识别成功！") print(f"完整文本:\n{result.full_text}") def example_3_engine_directly(): """示例 3: 直接使用 OCR 引擎""" print("\n示例 3: 直接使用 OCR 引擎") print("-" * 50) # 创建引擎 config = { 'lang': 'ch', # 语言 'use_gpu': False, # 是否使用 GPU 'show_log': False # 是否显示日志 } engine = PaddleOCREngine(config) # 识别图片 result = engine.recognize( image="path/to/your/image.png", preprocess=False ) if result.success: print(f"识别成功！") print(f"完整文本:\n{result.full_text}") # 遍历每一行 for line_result in result.results: print(f"行 {line_result.line_index}: {line_result.text} (置信度: {line_result.confidence:.2f})") def example_4_batch_images(): """示例 4: 批量处理多张图片""" print("\n示例 4: 批量处理多张图片") print("-" * 50) image_list = [ "path/to/image1.png", "path/to/image2.png", "path/to/image3.png" ] engine = PaddleOCREngine({'lang': 'ch'}) for i, image_path in enumerate(image_list, 1): print(f"\n处理图片 {i}: {image_path}") result = engine.recognize(image_path) if result.success: print(f" 识别成功，置信度: {result.total_confidence:.2f}") print(f" 文本预览: {result.full_text[:100]}...") else: print(f" 识别失败: {result.error_message}") def example_5_image_preprocess(): """示例 5: 图像预处理（增强识别效果）""" print("\n示例 5: 图像预处理") print("-" * 50) # 预处理并保存 processed = preprocess_image( image_path="path/to/input.png", output_path="path/to/output_processed.png", resize=True, # 调整大小 enhance_contrast=True, # 增强对比度 enhance_sharpness=True, # 增强锐度 denoise=False, # 不去噪 binarize=False # 不二值化 ) print(f"预处理完成，图像尺寸: {processed.size}") # 然后对预处理后的图片进行 OCR result = recognize_text( image=processed, # 可以传入 PIL Image mode="local", lang="ch" ) if result.success: print(f"识别文本: {result.full_text}") def example_6_multilanguage(): """示例 6: 多语言识别""" print("\n示例 6: 多语言识别") print("-" * 50) # 中文 result_ch = recognize_text( image="path/to/chinese_image.png", lang="ch" # 中文 ) print(f"中文识别置信度: {result_ch.total_confidence:.2f}") # 英文 result_en = recognize_text( image="path/to/english_image.png", lang="en" # 英文 ) print(f"英文识别置信度: {result_en.total_confidence:.2f}") # 中英混合 result_mix = recognize_text( image="path/to/mixed_image.png", lang="chinese_chinese" # 中英混合 ) print(f"混合识别置信度: {result_mix.total_confidence:.2f}") def example_7_cloud_ocr(): """示例 7: 云端 OCR（需要配置）""" print("\n示例 7: 云端 OCR") print("-" * 50) # 配置云端 OCR config = { 'api_endpoint': 'https://api.example.com/ocr', 'api_key': 'your_api_key_here', 'provider': 'custom', 'timeout': 30 } engine = CloudOCREngine(config) # 注意：云端 OCR 需要根据具体 API 实现 _send_request 方法 result = engine.recognize("path/to/image.png") if result.success: print(f"识别成功: {result.full_text}") else: print(f"云端 OCR 尚未实现: {result.error_message}") def example_8_factory_pattern(): """示例 8: 使用工厂模式创建引擎""" print("\n示例 8: 使用工厂模式创建引擎") print("-" * 50) # 创建本地引擎 local_engine = OCRFactory.create_engine( mode="local", config={'lang': 'ch'} ) print(f"本地引擎类型: {type(local_engine).__name__}") # 创建云端引擎 cloud_engine = OCRFactory.create_engine( mode="cloud", config={'api_endpoint': 'https://api.example.com/ocr'} ) print(f"云端引擎类型: {type(cloud_engine).__name__}") def example_9_detailed_result(): """示例 9: 处理详细识别结果""" print("\n示例 9: 处理详细识别结果") print("-" * 50) result = recognize_text( image="path/to/image.png", mode="local", lang="ch" ) if result.success: # 遍历每一行结果 for line_result in result.results: print(f"\n行 {line_result.line_index}:") print(f" 文本: {line_result.text}") print(f" 置信度: {line_result.confidence:.2f}") # 如果有坐标信息 if line_result.bbox: print(f" 坐标: {line_result.bbox}") # 统计信息 total_chars = sum(len(r.text) for r in result.results) avg_confidence = sum(r.confidence for r in result.results) / len(result.results) print(f"\n统计:") print(f" 总行数: {len(result.results)}") print(f" 总字符数: {total_chars}") print(f" 平均置信度: {avg_confidence:.2f}") def example_10_pil_image_input(): """示例 10: 使用 PIL Image 作为输入""" print("\n示例 10: 使用 PIL Image 作为输入") print("-" * 50) from PIL import Image # 加载图像 pil_image = Image.open("path/to/image.png") # 裁剪感兴趣区域 cropped = pil_image.crop((100, 100, 500, 300)) # 直接识别 PIL Image result = recognize_text( image=cropped, # 直接传入 PIL Image 对象 mode="local", lang="ch" ) if result.success: print(f"识别结果: {result.full_text}") if __name__ == '__main__': print("OCR 模块使用示例") print("=" * 50) print("\n注意：运行这些示例前，请确保:") print("1. 安装依赖: pip install paddleocr paddlepaddle") print("2. 将示例中的 'path/to/image.png' 替换为实际图片路径") print("=" * 50) # 取消注释想要运行的示例 # example_1_quick_recognize() # example_2_with_preprocess() # example_3_engine_directly() # example_4_batch_images() # example_5_image_preprocess() # example_6_multilanguage() # example_7_cloud_ocr() # example_8_factory_pattern() # example_9_detailed_result() # example_10_pil_image_input()