Python处理Google Vision API OCR提取行文本与bounding box坐标

完整实现代码 #

from google.cloud import vision
from google.cloud.vision import types
import json
import io
from google.protobuf.json_format import MessageToJson

# other code…
client = vision.ImageAnnotatorClient()
file = "你的图片路径" # 替换为实际图片路径

with io.open(file, 'rb') as image_file:
        content = image_file.read()
        image = types.Image(content=content)
        response = client.document_text_detection(image=image)
        document = response.full_text_annotation

        serialized = MessageToJson(document)
        data = json.loads(serialized)

        symbols = []
        for page in data['pages']:
            for block in page['blocks']:
                for paragraph in block['paragraphs']:
                    for word in paragraph['words']:
                        for symbol in word['symbols']:
                            symbols.append(symbol)

        break_values = ['EOL_SURE_SPACE', 'HYPHEN', 'LINE_BREAK']
        space_values = ['UNKNOWN', 'SPACE', 'SURE_SPACE']

        # 初始化临时变量缓存当前行数据
        current_line = {
            "content": "",
            "first_symbol": None,
            "last_symbol": None
        }
        lines = []

        for symbol in symbols:
            # 记录当前行第一个符号用于取左上角坐标
            if not current_line["first_symbol"]:
                current_line["first_symbol"] = symbol
            # 拼接符号文本
            current_line["content"] += symbol["text"]
            current_line["last_symbol"] = symbol

            # 处理分隔符逻辑
            break_type = None
            if "property" in symbol and "detectedBreak" in symbol["property"]:
                break_type = symbol["property"]["detectedBreak"]["type"]
            
            if break_type in space_values:
                current_line["content"] += " "
            elif break_type in break_values:
                # 遇到换行，计算当前行的坐标、宽高
                first_vertex = current_line["first_symbol"]["boundingBox"]["vertices"][0]
                last_vertex = current_line["last_symbol"]["boundingBox"]["vertices"][2]
                line_x = first_vertex["x"]
                line_y = first_vertex["y"]
                line_w = last_vertex["x"] - line_x
                line_h = last_vertex["y"] - line_y
                # 存入结果列表
                lines.append({
                    "Content": current_line["content"].strip(),
                    "X": line_x,
                    "Y": line_y,
                    "H": line_h,
                    "W": line_w
                })
                # 重置临时变量
                current_line = {
                    "content": "",
                    "first_symbol": None,
                    "last_symbol": None
                }
        
        # 处理最后一行没有换行标记的剩余内容
        if current_line["content"].strip():
            first_vertex = current_line["first_symbol"]["boundingBox"]["vertices"][0]
            last_vertex = current_line["last_symbol"]["boundingBox"]["vertices"][2]
            line_x = first_vertex["x"]
            line_y = first_vertex["y"]
            line_w = last_vertex["x"] - line_x
            line_h = last_vertex["y"] - line_y
            lines.append({
                "Content": current_line["content"].strip(),
                "X": line_x,
                "Y": line_y,
                "H": line_h,
                "W": line_w
            })

# 输出Markdown格式表格
print("| Content | X | Y | H | W |")
print("| --- | --- | --- | --- | --- |")
for line in lines:
    print(f"| {line['Content']} | {line['X']} | {line['Y']} | {line['H']} | {line['W']} |")

逻辑说明 #

1. 用临时变量缓存当前行的文本内容、行首符号、行尾符号，避免频繁读写结果列表
2. 遍历每个符号时先拼接文本，再判断符号后的分隔符类型，空格类型直接加空格，换行类型则计算当前行的坐标和宽高存入结果列表，随后重置临时变量
3. 遍历结束后额外处理最后一行没有换行标记的剩余内容，避免遗漏
4. 最终直接按要求打印Markdown格式的表格，也可根据需求将lines列表导出为CSV、Excel等其他表格格式

内容的提问来源于stack exchange，提问作者paratext