图片翻译Image Translation

上传图片，通过高精度 OCR 识别图中文字并翻译为目标语言。支持多种图片格式、多语言文字识别、术语库集成以及指定翻译风格。Upload images for OCR-based text recognition and translation into the target language. Supports multiple image formats, multi-language recognition, glossary integration, and custom translation styles.

快速概览
端点EndpointPOST /v1/image/translate
认证方式AuthBearer Token
输入方式Input文件上传 (multipart) 或 图片 URL (JSON)File upload (multipart) or image URL (JSON)
图片上限Max Size20 MB
输出模式Output Mode纯文本 / 按区域含坐标Plain text / By region with coordinates
支持语言Languages50+ 语言50+ languages

请求端点

Endpoint

POST/v1/image/translate

认证

Authentication

Authorization: Bearer {access_token}

请求头

Request Headers

请求头Header必填Required说明Description
AuthorizationrequiredBearer 认证，格式：Bearer {access_token}Bearer authentication, format: Bearer {access_token}
Content-Typerequired文件上传时使用 multipart/form-data；JSON 模式使用 application/jsonUse multipart/form-data for file upload; application/json for JSON mode

请求参数

Request Parameters

参数Parameter类型Type必填Required说明Description
imagefile / stringrequired图片文件（multipart 上传）或图片 URL（JSON 模式）。最大 20MB。建议分辨率不低于 200 DPI 以获得最佳识别效果。Image file (multipart) or image URL (JSON). Max 20MB. Recommended at least 200 DPI for best recognition.
target_langstringrequired目标翻译语言代码，详见语种列表Target language code, see Language List
source_langstringoptional图片中文字语言代码。默认自动检测。当已知语言时建议显式指定以提高识别准确度。Source language code. Auto-detected by default. Specifying it improves accuracy when known.
output_typestringoptionaltext — 仅返回纯文本翻译结果（不含位置信息）；
regions — 按区域返回，每块文本含坐标 bbox 及原文与译文。默认 text		text — plain text only (no position info);
regions — per region with bbox, original and translated text. Default text
formalitystringoptional翻译语气：default（默认）、more（更正式）、less（更随意）。仅特定语言支持，见下方表单性支持表。Formality: default, more (formal), less (casual). Only supported for certain languages, see formality table below.
glossary_idstringoptional关联术语库 ID，翻译时优先采用术语库中的定义Glossary ID for custom term definitions during translation
preserve_formattingbooleanoptional是否保留原文中的换行、空格等格式。默认 falsePreserve line breaks and whitespace. Default false
model_typestringoptional模型选择：standard（标准，默认）、prefer_quality（高质量，适用于复杂排版）Model: standard (default), prefer_quality (higher quality for complex layouts)
show_billed_charactersbooleanoptional是否在响应中返回计费字符数。默认 falseReturn billed character count in response. Default false

表单性支持

Formality Support

以下语言在图片翻译中支持 formality 参数，可控制翻译结果的语气正式程度：The following languages support the formality parameter in image translation to control tone:

语言Language语言代码Code
德语GermanDE
法语FrenchFR
意大利语ItalianIT
西班牙语SpanishES
葡萄牙语PortuguesePT
俄语RussianRU
日语JapaneseJA
韩语KoreanKO
荷兰语DutchNL
波兰语PolishPL

支持图片格式

Supported Image Formats

格式FormatMIME 类型MIME Type说明Notes
JPEG / JPGimage/jpeg最常用格式，推荐质量 80% 以上Most common, recommended quality > 80%
PNGimage/png支持透明通道，适合截图和图标Supports alpha channel, ideal for screenshots and icons
WebPimage/webp高压缩率，同样质量下文件更小High compression, smaller file size at same quality
BMPimage/bmp无压缩位图，文件较大但无质量损失Uncompressed bitmap, larger files but lossless
TIFFimage/tiff常用于文档扫描，支持多页（仅第一页被处理）Common for scanned documents, multi-page (first page only)
PDF（单页）application/pdf仅支持单页 PDF，多页请使用文档翻译Single-page PDF only; use Document Translation for multi-page

请求示例

Request Examples

cURL — 文件上传

cURL — File Upload

curl -X POST https://api.itranslator.cc/v1/image/translate \
  -H "Authorization: Bearer YOUR_ACCESS_TOKEN" \
  -F "image=@menu.jpg" \
  -F "target_lang=en" \
  -F "source_lang=ja" \
  -F "output_type=regions" \
  -F "formality=more"

cURL — URL 模式

cURL — URL Mode

curl -X POST https://api.itranslator.cc/v1/image/translate \
  -H "Authorization: Bearer YOUR_ACCESS_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "image": "https://example.com/menu.jpg",
    "target_lang": "en",
    "source_lang": "ja",
    "output_type": "text",
    "glossary_id": "gl_abc123"
  }'

Python — 文件上传

Python — File Upload

import requests

url = "https://api.itranslator.cc/v1/image/translate"
headers = {"Authorization": f"Bearer {token}"}

with open("menu.jpg", "rb") as img:
    resp = requests.post(
        url,
        headers=headers,
        files={"image": img},
        data={
            "target_lang": "en",
            "source_lang": "ja",
            "output_type": "regions",
            "formality": "more"
        }
    )

result = resp.json()
for region in result["regions"]:
    print(f"{region['source_text']} → {region['translation']}")

Python — URL 模式

Python — URL Mode

import requests

resp = requests.post(
    "https://api.itranslator.cc/v1/image/translate",
    headers={
        "Authorization": f"Bearer {token}",
        "Content-Type": "application/json"
    },
    json={
        "image": "https://example.com/sign.jpg",
        "target_lang": "zh",
        "source_lang": "en",
        "output_type": "text",
        "show_billed_characters": True
    }
)
print(resp.json())

JavaScript (Node.js)

JavaScript (Node.js)

const fs = require("fs");

const form = new FormData();
form.append("image", fs.createReadStream("./menu.jpg"));
form.append("target_lang", "en");
form.append("output_type", "regions");

const resp = await fetch(
  "https://api.itranslator.cc/v1/image/translate",
  {
    method: "POST",
    headers: { Authorization: `Bearer ${token}` },
    body: form
  }
);
const data = await resp.json();
console.log(data);

Java

Java

import java.net.URI;
import java.net.http.*;

// 使用 URL 模式
HttpClient client = HttpClient.newHttpClient();
String json = """
  {"image":"https://example.com/menu.jpg","target_lang":"en","output_type":"text"}
  """;
HttpRequest req = HttpRequest.newBuilder()
  .uri(URI.create("https://api.itranslator.cc/v1/image/translate"))
  .header("Authorization", "Bearer " + token)
  .header("Content-Type", "application/json")
  .POST(HttpRequest.BodyPublishers.ofString(json))
  .build();
HttpResponse<String> resp = client.send(req,
  HttpResponse.BodyHandlers.ofString());
System.out.println(resp.body());

响应字段说明

Response Fields

字段Field类型Type说明Description
source_langstring检测或指定的源语言代码Detected or specified source language code
target_langstring目标翻译语言代码Target language code
full_textstring完整翻译后的纯文本（output_type=text 时为主要输出）Full translated text (main output when output_type=text)
regionsarray区域数组（仅 output_type=regions），每项含 bbox、source_text、translation、confidenceRegion array (output_type=regions only), each with bbox, source_text, translation, confidence
billed_charactersinteger计费字符数（仅 show_billed_characters=true 时返回）Billed character count (only when show_billed_characters=true)
detected_source_langstring自动检测到的源语言（source_lang 未指定时返回）Auto-detected source language (returned when source_lang not specified)
model_type_usedstring实际使用的翻译模型The translation model actually used
processing_time_msinteger处理耗时（毫秒）Processing time in milliseconds

响应示例

Response Examples

示例 1：output_type=text（默认）

Example 1: output_type=text (default)

{
  "source_lang": "ja",
  "target_lang": "en",
  "detected_source_lang": "ja",
  "full_text": "Assorted Sashimi $28.00\nToday's Special - Grilled Salmon $18.50",
  "billed_characters": 48,
  "model_type_used": "standard",
  "processing_time_ms": 320
}

示例 2：output_type=regions（含位置信息）

Example 2: output_type=regions (with position)

{
  "source_lang": "ja",
  "target_lang": "en",
  "regions": [
    {
      "bbox": { "x": 120, "y": 45, "w": 200, "h": 30 },
      "source_text": "刺身盛り合わせ",
      "translation": "Assorted Sashimi",
      "confidence": 0.98
    },
    {
      "bbox": { "x": 380, "y": 45, "w": 160, "h": 30 },
      "source_text": "$28.00",
      "translation": "$28.00",
      "confidence": 1.00
    },
    {
      "bbox": { "x": 120, "y": 85, "w": 300, "h": 30 },
      "source_text": "本日のおすすめ — 焼き鮭",
      "translation": "Today's Special — Grilled Salmon",
      "confidence": 0.95
    },
    {
      "bbox": { "x": 420, "y": 85, "w": 120, "h": 30 },
      "source_text": "$18.50",
      "translation": "$18.50",
      "confidence": 1.00
    }
  ],
  "full_text": "Assorted Sashimi $28.00\nToday's Special — Grilled Salmon $18.50",
  "processing_time_ms": 380
}

示例 3：含术语库 + 表单性

Example 3: With Glossary + Formality

{
  "source_lang": "zh",
  "target_lang": "ja",
  "regions": [
    {
      "bbox": { "x": 50, "y": 30, "w": 180, "h": 28 },
      "source_text": "欢迎光临",
      "translation": "いらっしゃいませ",
      "confidence": 0.99
    },
    {
      "bbox": { "x": 50, "y": 70, "w": 220, "h": 28 },
      "source_text": "新品上市 — 抹茶拿铁",
      "translation": "新発売 — 抹茶ラテ（※用語集適用）",
      "confidence": 0.97
    }
  ],
  "full_text": "いらっしゃいませ\n新発売 — 抹茶ラテ（※用語集適用）",
  "model_type_used": "prefer_quality",
  "processing_time_ms": 520
}

错误码

Error Codes

Status错误码Code说明Description
400UNSUPPORTED_IMAGE不支持的图片格式或图片已损坏Unsupported or corrupted image format
400IMAGE_TOO_LARGE图片超过 20MB 限制Image exceeds 20MB limit
400NO_TEXT_DETECTED图片中未检测到可识别的文字No recognizable text detected in image
400INVALID_IMAGE_URL图片 URL 无效、无法访问或超时Invalid, inaccessible or timed-out image URL
400INVALID_LANGUAGE不支持的语言代码Unsupported language code
401UNAUTHORIZED缺少或无效的 Access TokenMissing or invalid access token
403FORBIDDEN无权访问该资源或配额已耗尽Access denied or quota exceeded
413PAYLOAD_TOO_LARGE请求体超过服务器限制Request body too large
429RATE_LIMITED请求频率超限，请降低并发数Rate limit exceeded, reduce concurrency
456GLOSSARY_NOT_FOUND指定的术语库不存在或已被删除Specified glossary not found or deleted
500INTERNAL_ERROR服务器内部错误，请稍后重试Internal server error, try again later
503SERVICE_UNAVAILABLE服务暂时不可用，请稍后重试Service temporarily unavailable

最佳实践

Best Practices

  1. 图片质量：确保图片清晰，分辨率不低于 200 DPI。模糊或低光照的图片会显著降低识别准确率。
  2. Image Quality: Ensure images are clear with at least 200 DPI resolution. Blurry or low-light images significantly reduce recognition accuracy.
  3. 指定源语言：当已知图片文字语言时，建议通过 source_lang 显式指定，避免自动检测带来的延迟和误判。
  4. Specify Source Language: Always specify source_lang when known to avoid auto-detection delay and misidentification.
  5. 选择合适的 output_type：如果只需要翻译文本，使用 text 模式更快；如果需要定位还原到图片，使用 regions 模式获取坐标。
  6. Choose Output Type: Use text mode for plain translated text (faster); use regions mode when you need coordinates for overlay.
  7. 结合术语库：对特定领域（医疗、法律、游戏等）的图片翻译，需要传入 glossary_id 以确保专业术语的翻译一致性。
  8. Use Glossary: For domain-specific images (medical, legal, gaming, etc.), pass glossary_id to ensure consistent professional terminology.
  9. 大图预处理：超过 10MB 的图片建议先压缩到合适大小再上传，可显著减少网络传输时间和请求延迟。
  10. Pre-process Large Images: Compress images over 10MB before upload to reduce network latency and request time.
  11. 错误重试：对于 429（频率限制）和 5xx（服务器错误），建议实现指数退避重试策略。
  12. Retry Strategy: Implement exponential backoff for 429 (rate limit) and 5xx (server errors) responses.

常见应用场景

Common Use Cases

场景Scenario说明Description
菜单翻译Menu Translation拍摄菜单即可获得翻译，建议使用 regions 模式将翻译叠加到图片原位Snap a menu for translation; use regions mode to overlay translations in place
路标/指示牌Signage & Signboards出国旅行时拍摄路标、指示牌快速获取翻译Quick translation of street signs and direction boards while traveling
产品标签Product Labels跨境电商场景下翻译商品成分表、说明书标签，可配合术语库使用Translate ingredient lists and product labels for e-commerce, works with glossary
漫画/图文Comics & Manga翻译漫画中的对话气泡，regions 模式下可按区域逐条获取Translate speech bubbles in comics; regions provide text by bubble
截屏翻译Screenshot Translation翻译应用、游戏或网页截图中出现的文字，适用于本地化测试Translate text in app, game or web screenshots for localization testing

与 OCR 接口的区别

Difference from OCR API

图片翻译 APIImage Translate APIOCR API
功能Function识别 + 翻译Recognize + Translate仅识别文本，不做翻译Text recognition only, no translation
输出Output翻译后的目标语言文本Translated text in target language图片中的原始文字Original text from the image
适用Use Case直接需要翻译结果的场景When translated output is needed directly需要原始文本做后续处理（NLP、搜索、归档等）When original text is needed for downstream processing (NLP, search, archiving)
图片上限Max Size20 MB10 MB
使用说明
  • 文件上传使用 multipart/form-data 编码，Content-Type 请勿手动设置，让 HTTP 客户端自动生成。
  • Use multipart/form-data encoding; let the HTTP client auto-generate Content-Type.
  • 大文件建议先压缩或使用异步任务模式，避免请求超时。单次请求超时时间为 30 秒。
  • For large files, compress or use async task mode to avoid timeouts. Request timeout is 30 seconds.
  • 请勿在客户端代码中暴露 Access Token，建议通过后端代理调用。
  • Do not expose your Access Token in client-side code; use a backend proxy.
  • 如果图片包含密集文字（如文档扫描件），建议使用 model_type=prefer_quality 以获得更好效果。
  • For dense text images (e.g., scanned documents), use model_type=prefer_quality for better results.
← 语音翻译 视频翻译 →