Camera OCR

Performs OCR on camera photos with advanced image enhancement and preprocessing using ABBYY FineReader Engine.

Common Properties

• Name - The custom name of the node.
• Color - The custom color of the node.
• Delay Before (sec) - Waits in seconds before executing the node.
• Delay After (sec) - Waits in seconds after executing node.
• Continue On Error - Automation will continue regardless of any error. The default value is false.

Inputs

• Path - Path to the input image file captured from camera or smartphone.
• Out Path - Path where the processed output image will be saved.

Options

• Language - Language for OCR recognition (default: English). Set to "_" to skip text recognition and only perform image enhancement.
• Noise Model - Noise reduction model to apply (default: none "_"). Options:
  • None (_) - Skip noise removal
  • NMM_NoisyImage - For noisy or grainy images
  • NMM_BlurredImage - For blurred images
  • NMM_NoisyBlurredImage - For both noisy and blurred images
• Crop Image - Whether to automatically crop the image to content area (default: false).
• Correct Orientation - Whether to automatically detect and correct image orientation (default: true).
• Correct Geometrical Distortions - Whether to correct perspective and geometrical distortions (default: true).
• Deskew Image - Whether to straighten skewed images (default: true).
• Has Large Characters - Whether the image contains large font characters to optimize recognition (default: false).
• Remove Motion Blur - Whether to reduce motion blur from camera movement (default: true).
• Make White Background - Whether to normalize background to white color (default: true).

Outputs

• Statistics - Recognition statistics object containing processing time, quality metrics, and processed images.

How It Works

The Camera OCR node enhances camera-captured images and performs OCR. When executed, the node:

1. Validates the input file exists
2. Configures processing settings based on options
3. Warms up the processor engine
4. Processes the image with selected enhancements:
   • Crop to content area
   • Correct orientation
   • Fix geometric distortions
   • Deskew the image
   • Remove noise
   • Reduce motion blur
   • Normalize background
5. Optionally performs OCR if language is specified
6. Saves enhanced image(s) to output path
7. Returns processing statistics

Requirements

• Valid ABBYY FineReader Engine installation
• Valid ABBYY license
• Input image file from camera or smartphone
• Output directory must exist and be writable

Error Handling

The node will return specific errors in the following cases:

• ErrNotFound - Input image file not found. Error message includes the path.

Usage Example

Scenario: Enhance and OCR a smartphone photo of a document

Camera OCR node:
- Path: "C:/photos/receipt.jpg"
- Out Path: "C:/enhanced/receipt_clean.jpg"
- Language: English
- Noise Model: NMM_NoisyImage
- Crop Image: true
- Correct Orientation: true
- Correct Geometrical Distortions: true
- Deskew Image: true
- Has Large Characters: false
- Remove Motion Blur: true
- Make White Background: true

Scenario: Only enhance image without OCR

Camera OCR node:
- Path: "C:/photos/document.jpg"
- Out Path: "C:/enhanced/document_clean.jpg"
- Language: _ (skip OCR)
- Correct Orientation: true
- Deskew Image: true
- Remove Motion Blur: true
- Make White Background: true

Scenario: Process large text signage photo

Camera OCR node:
- Path: "C:/photos/sign.jpg"
- Out Path: "C:/processed/sign.jpg"
- Language: English
- Has Large Characters: true
- Correct Geometrical Distortions: true
- Remove Motion Blur: true

Common Use Cases

• Mobile Document Capture - Process documents photographed with smartphones
• Receipt Scanning - Enhance and extract text from receipt photos
• Business Card OCR - Clean up and read business card photos
• Form Processing - Process photos of filled forms
• Sign Reading - Extract text from photographed signs or labels
• Whiteboard Capture - Clean up whiteboard photos for OCR
• Book Scanning - Process book page photos
• License Plate Recognition - Enhance photos for plate reading (with large characters)

Image Enhancements

Crop Image

• Automatically detects document boundaries
• Removes unnecessary background
• Focuses processing on document area
• Improves OCR accuracy

Correct Orientation

• Detects if image is rotated
• Automatically rotates to correct orientation
• Critical for handheld photos
• Works with 90, 180, 270 degree rotations

Correct Geometrical Distortions

• Fixes perspective distortion from angled photos
• Corrects trapezoid effects
• Straightens curved text lines
• Essential for non-perpendicular captures

Deskew Image

• Straightens skewed images
• Corrects rotational misalignment
• Aligns text to horizontal
• Improves OCR line detection

Noise Removal

• Reduces grain and noise
• Improves text clarity
• Multiple models for different noise types
• Select model matching noise characteristics

Remove Motion Blur

• Reduces blur from camera movement
• Sharpens image details
• Critical for handheld photos
• Improves character recognition

Make White Background

• Normalizes background color
• Increases contrast
• Removes color casts
• Creates clean, scanner-like output

Output Statistics

The statistics object includes:

• Processing Time - Total time for image processing
• Quality Metrics - Image quality scores
• Enhanced Images - Array of processed image objects
• Recognition Results - OCR results if language specified

Multiple output images may be generated if document splitting occurs.

Tips and Best Practices

• When to Use:
  • Photos taken with smartphones or tablets
  • Handheld camera captures
  • Non-professional image sources
  • Images with motion blur or poor alignment
  • Photos of documents on surfaces (not scanned)
• Language Selection:
  • Specify language for OCR
  • Use "_" to only enhance without OCR
  • Combine enhancement with separate OCR for flexibility
• Noise Model:
  • Choose based on image characteristics
  • NMM_NoisyImage: Low light, high ISO photos
  • NMM_BlurredImage: Motion blur, out of focus
  • NMM_NoisyBlurredImage: Both issues present
  • None: Clean, well-lit photos
• Enhancement Options:
  • Enable all for unknown image quality
  • Disable specific options if not needed
  • Each enhancement adds processing time
  • More enhancements = better quality but slower
• Large Characters:
  • Enable for signs, headlines, posters
  • Disable for normal document text
  • Optimizes recognition for font size
  • Improves accuracy for large text
• Image Quality:
  • Higher resolution = better results
  • Good lighting improves all enhancements
  • Avoid extreme angles (over 30 degrees)
  • Keep camera steady for less blur
• Output Path:
  • Multiple images may be created (numbered)
  • Format: basename.jpg, basename_2.jpg, etc.
  • Ensure sufficient disk space
  • Output is always enhanced version
• Performance:
  • Enhancement takes time (3-10 seconds typical)
  • More corrections = longer processing
  • Image size affects processing time
  • Warming up engine is one-time overhead
• Best Capture Practices:
  • Use good lighting
  • Hold camera steady
  • Capture perpendicular to document
  • Include some margin around document
  • Avoid shadows on document
  • Use highest camera resolution
• Comparison with Scanners:
  • Camera OCR bridges gap between camera and scanner
  • Enhancements make photos scanner-quality
  • Ideal for mobile workflows
  • More forgiving than raw OCR
• Workflow Integration:
  • Use Camera OCR as preprocessing step
  • Follow with standard Process Document if needed
  • Extract statistics for quality monitoring
  • Archive both original and enhanced images
• Error Handling:
  • Enable Continue On Error for batches
  • Check statistics for quality issues
  • Review low-quality images manually
  • Recapture if enhancement insufficient