> ## Documentation Index
> Fetch the complete documentation index at: https://mintlify.com/microsoft/powertoys/llms.txt
> Use this file to discover all available pages before exploring further.

# Text Extractor

> Extract text from anywhere on your screen using OCR (Optical Character Recognition) technology.

Text Extractor (PowerOCR) uses Windows OCR to extract text from any region of your screen. Copy text from images, videos, dialogs, or any application where text selection isn't possible.

## Activation

<Steps>
  <Step title="Activate OCR Mode">
    Press `Win + Shift + T` to activate Text Extractor.
  </Step>

  <Step title="Select Region">
    Click and drag to select the screen region containing text.
  </Step>

  <Step title="Extract Text">
    Release mouse button - text is automatically extracted and copied to clipboard.
  </Step>

  <Step title="Paste Text">
    Press `Ctrl + V` to paste the extracted text anywhere.
  </Step>
</Steps>

<Note>
  Customize the activation shortcut in PowerToys Settings > Text Extractor.
</Note>

## OCR Technology

### Windows OCR Engine

Text Extractor uses Windows' built-in OCR engine (Windows.Media.Ocr) for text recognition.

From source `/src/modules/PowerOCR/PowerOCR/Helpers/OcrExtensions.cs:90-106`:

<CodeGroup>
  ```csharp OCR Implementation theme={null}
  internal static async Task<OcrResult> GetOcrResultFromImageAsync(Bitmap bmp, Language language)
  {
      await using MemoryStream memoryStream = new();
      using WrappingStream wrappingStream = new(memoryStream);
      
      bmp.Save(wrappingStream, ImageFormat.Bmp);
      wrappingStream.Position = 0;
      
      BitmapDecoder bmpDecoder = await BitmapDecoder.CreateAsync(wrappingStream.AsRandomAccessStream());
      SoftwareBitmap softwareBmp = await bmpDecoder.GetSoftwareBitmapAsync();
      
      OcrEngine ocrEngine = OcrEngine.TryCreateFromLanguage(language);
      return await ocrEngine.RecognizeAsync(softwareBmp);
  }
  ```
</CodeGroup>

### Supported Languages

Text Extractor supports all languages installed in Windows:

<Accordion title="Language Support">
  OCR works with:

  * **Latin scripts:** English, Spanish, French, German, Italian, Portuguese, etc.
  * **Cyrillic scripts:** Russian, Ukrainian, Serbian, Bulgarian, etc.
  * **Asian languages:** Chinese (Simplified & Traditional), Japanese, Korean
  * **Other scripts:** Arabic, Hebrew, Greek, Thai, etc.

  **Note:** Language must be installed in Windows Settings > Time & Language > Language.
</Accordion>

From source `/src/modules/PowerOCR/PowerOCR/Helpers/LanguageHelper.cs`

## Features

### Text Recognition Modes

<Tabs>
  <Tab title="Standard OCR">
    Default mode extracts text line-by-line:

    * Preserves line breaks
    * Maintains text order
    * Handles multi-line paragraphs
    * Good for most text extraction
  </Tab>

  <Tab title="Table Mode">
    Extracts text organized in tables:

    * Detects table structure
    * Preserves column alignment
    * Maintains row/column relationships
    * Useful for spreadsheets and data

    Implementation: `/src/modules/PowerOCR/PowerOCR/Helpers/OcrExtensions.cs:66-88`
  </Tab>
</Tabs>

### Language-Aware Text Processing

Text Extractor intelligently handles different language types:

<Accordion title="Space-Joining Languages">
  From source `/src/modules/PowerOCR/PowerOCR/Helpers/OcrExtensions.cs:25-64`:

  **Space-joining languages** (English, Spanish, etc.):

  * Automatically add spaces between words
  * Preserve word boundaries
  * Handle punctuation correctly

  **Non-space-joining languages** (Chinese, Japanese):

  * No automatic spacing between characters
  * Detect when to insert spaces (for mixed content)
  * Handle CJK characters as single-character words

  <CodeGroup>
    ```csharp Language Detection theme={null}
    bool isSpaceJoiningOCRLang = LanguageHelper.IsLanguageSpaceJoining(language);

    if (isSpaceJoiningOCRLang) {
        text.AppendLine(ocrLine.Text);
    } else {
        // Complex logic to handle CJK + Latin mixed text
        Regex regexSpaceJoiningWord = new(@"(^[\p{L}-[\p{Lo}]]|\p{Nd}$)|.{2,}");
        // Insert spaces only where appropriate
    }
    ```
  </CodeGroup>
</Accordion>

### Image Scaling

Automatic image scaling for better OCR accuracy:

<CodeGroup>
  ```csharp Image Scaling theme={null}
  bool scaleBMP = true;

  if (bmp.Width * 1.5 > OcrEngine.MaxImageDimension) {
      scaleBMP = false;  // Image already large enough
  }

  // Scale to 1.5x for better recognition (if within limits)
  using Bitmap scaledBitmap = scaleBMP ? 
      ImageMethods.ScaleBitmapUniform(bmp, 1.5) : 
      ImageMethods.ScaleBitmapUniform(bmp, 1.0);
  ```
</CodeGroup>

From `/src/modules/PowerOCR/PowerOCR/Helpers/OcrExtensions.cs:75-82`

<Tip>
  Text Extractor automatically scales images up to 1.5x for improved OCR accuracy while respecting the OCR engine's maximum dimension limit.
</Tip>

## Use Cases

<CardGroup cols={2}>
  <Card title="Copy Text from Images" icon="image">
    Extract text from screenshots, photos, scanned documents, or any image with text.
  </Card>

  <Card title="Extract from Videos" icon="video">
    Capture text from paused video frames, subtitles, or video overlays.
  </Card>

  <Card title="Legacy Applications" icon="window-restore">
    Copy text from applications that don't support text selection or copying.
  </Card>

  <Card title="Error Messages" icon="triangle-exclamation">
    Capture and share error messages from dialog boxes or system alerts.
  </Card>
</CardGroup>

<AccordionGroup>
  <Accordion title="Development Workflows">
    **Extract Code from Screenshots:**

    * Copy code from tutorial screenshots
    * Extract code from Stack Overflow images
    * Capture configuration examples from documentation
    * Get code from video tutorials

    **Example:** Extract SQL query from database tool screenshot.
  </Accordion>

  <Accordion title="Data Entry">
    **Digitize Printed/Handwritten Text:**

    * Extract text from scanned forms
    * Digitize business cards
    * Copy data from PDF images
    * Extract information from screenshots

    **Example:** Extract address from scanned invoice image.
  </Accordion>

  <Accordion title="Research & Documentation">
    **Capture Reference Information:**

    * Extract quotes from e-books (images)
    * Copy data from research papers (PDFs)
    * Capture table data from reports
    * Save information from presentations

    **Example:** Extract table of statistics from presentation slide.
  </Accordion>

  <Accordion title="Accessibility">
    **Text-to-Speech Support:**

    * Extract text for screen readers
    * Copy text from inaccessible interfaces
    * Get text from images for translation
    * Support users with visual impairments

    **Example:** Extract text from image to read with text-to-speech.
  </Accordion>

  <Accordion title="Translation">
    **Multilingual Text Extraction:**

    * Extract foreign language text from images
    * Copy text for translation services
    * Capture text from international websites
    * Get text from foreign language videos

    **Example:** Extract Japanese text from game screenshot for translation.
  </Accordion>
</AccordionGroup>

## Configuration

### Activation Shortcut

<Steps>
  <Step title="Open Settings">
    Open PowerToys Settings > Text Extractor.
  </Step>

  <Step title="Change Shortcut">
    Click activation shortcut field and press your desired key combination (default: `Win + Shift + T`).
  </Step>

  <Step title="Test">
    Use new shortcut to verify it works.
  </Step>
</Steps>

### OCR Language

Select the language for text recognition:

<Steps>
  <Step title="Set Primary Language">
    Choose the language most commonly used in text you extract.
  </Step>

  <Step title="Add Languages">
    Install additional OCR languages in Windows Settings if needed:

    Settings > Time & Language > Language > Add a language
  </Step>

  <Step title="Switch Languages">
    Change language in Text Extractor settings before extracting text in different language.
  </Step>
</Steps>

<Warning>
  OCR accuracy depends on selecting the correct language. Using English OCR on Chinese text will produce poor results.
</Warning>

### Display Options

* Cursor style during selection
* Overlay color and opacity
* Selection border appearance
* Result preview window

## Advanced Features

### Table Recognition

Extract structured table data as formatted text:

<CodeGroup>
  ```csharp Table Extraction theme={null}
  public static async Task<string> GetRegionsTextAsTableAsync(
      OCROverlay passedWindow, 
      Rectangle regionScaled, 
      Language? language)
  {
      // Extract text
      OcrResult ocrResult = await GetOcrResultFromImageAsync(scaledBitmap, language);
      
      // Parse into table structure
      List<WordBorder> wordBorders = ResultTable.ParseOcrResultIntoWordBorders(
          ocrResult, dpiScale);
      
      // Format as table
      return ResultTable.GetWordsAsTable(
          wordBorders, dpiScale, isSpaceJoiningLang);
  }
  ```
</CodeGroup>

From `/src/modules/PowerOCR/PowerOCR/Helpers/OcrExtensions.cs:66-88`

### DPI-Aware Selection

Text Extractor is DPI-aware for accurate region selection on high-DPI displays:

```csharp theme={null}
DpiScale dpiScale = VisualTreeHelper.GetDpi(passedWindow);
```

Ensures pixel-perfect selection on 4K monitors and mixed-DPI setups.

### Image Processing Helpers

Utility functions for image manipulation:

Implementation: `/src/modules/PowerOCR/PowerOCR/Helpers/ImageMethods.cs`

* Bitmap scaling and uniform resizing
* Region capture from screen
* DPI-aware coordinate conversion
* Image format conversion

### Cursor Management

During text selection, cursor is confined to prevent accidental clicks outside overlay:

Implementation: `/src/modules/PowerOCR/PowerOCR/Helpers/CursorClipper.cs`

### String Processing

Post-processing for extracted text:

Implementation: `/src/modules/PowerOCR/PowerOCR/Helpers/StringHelpers.cs`

* Whitespace normalization
* Line break handling
* Special character cleanup
* Format preservation

## OCR Accuracy Tips

<AccordionGroup>
  <Accordion title="Improve Recognition Quality">
    **Best Practices:**

    1. **Select tightly around text** - Minimize background and non-text areas
    2. **Use high-resolution sources** - Zoom in or use higher quality images
    3. **Ensure good contrast** - Dark text on light background (or vice versa)
    4. **Avoid distorted text** - Skewed or rotated text reduces accuracy
    5. **Select correct language** - Match OCR language to text language
    6. **Clean text areas** - Remove visual noise or overlapping elements
  </Accordion>

  <Accordion title="Handle Difficult Text">
    **Challenging Scenarios:**

    * **Small text:** Zoom in or scale up the window before capturing
    * **Stylized fonts:** Simple fonts work better than decorative ones
    * **Backgrounds:** Solid backgrounds better than textured/busy ones
    * **Mixed languages:** Extract each language separately with correct language setting
    * **Handwriting:** Printed text works better; handwriting recognition is limited
  </Accordion>

  <Accordion title="Post-Processing**">
    After extraction:

    1. Review extracted text for errors
    2. Correct misrecognized characters (common: O/0, I/l/1, S/5)
    3. Fix spacing issues in mixed-language text
    4. Preserve original formatting where needed
    5. Use spell-check to catch obvious errors
  </Accordion>
</AccordionGroup>

## Keyboard Shortcuts

| Shortcut          | Action                            |
| ----------------- | --------------------------------- |
| `Win + Shift + T` | Activate Text Extractor (default) |
| `Esc`             | Cancel text extraction            |
| `Click + Drag`    | Select region to extract          |
| `Ctrl + V`        | Paste extracted text              |

## Troubleshooting

<AccordionGroup>
  <Accordion title="Text Extractor Not Activating">
    **Checklist:**

    * Ensure Text Extractor is enabled in PowerToys Settings
    * Verify activation shortcut isn't conflicting
    * Check PowerToys is running
    * Restart PowerToys if needed

    **Common conflicts:** Some applications use `Win + Shift + T` - choose different shortcut.
  </Accordion>

  <Accordion title="Poor OCR Accuracy">
    **Troubleshooting:**

    * Verify correct language is selected
    * Check text is clear and readable
    * Ensure sufficient contrast
    * Try zooming in on text before extraction
    * Avoid selecting too much non-text area
    * Check Windows OCR language pack is installed

    **Note:** OCR accuracy varies by font, size, and clarity.
  </Accordion>

  <Accordion title="No Text Detected">
    **Possible Causes:**

    * Selected area contains no text
    * Text is too small or blurry
    * Wrong language selected
    * Image contrast too low
    * Text is actually rendered graphics (not characters)

    **Solution:** Try larger selection area, zoom in, or adjust image source.
  </Accordion>

  <Accordion title="Clipboard Not Updated">
    **Issues:**

    * Another application may be blocking clipboard
    * Clipboard manager interference
    * No text was detected (check extraction was successful)

    **Fix:** Close clipboard managers temporarily, verify text was extracted.
  </Accordion>

  <Accordion title="Language Not Available">
    **Missing Language Pack:**

    1. Open Windows Settings
    2. Go to Time & Language > Language
    3. Click "Add a language"
    4. Select desired language
    5. Check "Optical character recognition" in optional features
    6. Install language pack
    7. Restart Text Extractor
  </Accordion>
</AccordionGroup>

## Performance Considerations

<Accordion title="OCR Performance">
  **Processing Time:**

  * Small text regions: \< 1 second
  * Medium regions: 1-3 seconds
  * Large regions: 3-10 seconds
  * Very large regions: May timeout or fail

  **Recommendations:**

  * Select only necessary text areas
  * Break large extractions into smaller regions
  * Be patient with complex multi-language text
  * Close other heavy applications during large extractions
</Accordion>

## Accessibility Features

<Accordion title="Accessibility Support">
  Text Extractor improves accessibility:

  * **Screen Reader Compatible:** Extracted text can be read by screen readers
  * **Keyboard Navigation:** Full keyboard control during selection
  * **High Contrast Support:** Works with Windows high contrast themes
  * **Magnification:** Works with Windows Magnifier
  * **Text-to-Speech:** Extract text for speech synthesis
</Accordion>

## Source Code

Location: `/src/modules/PowerOCR/PowerOCR/`

* OCR extensions: `Helpers/OcrExtensions.cs`
* Language support: `Helpers/LanguageHelper.cs`
* Image processing: `Helpers/ImageMethods.cs`
* String utilities: `Helpers/StringHelpers.cs`
* Cursor management: `Helpers/CursorClipper.cs`
* Throttled actions: `Helpers/ThrottledActionInvoker.cs`
* OS interop: `Helpers/OSInterop.cs`
* Main application: `App.xaml.cs`