How to Detect Page Rotation in C# with IronOCR

What Do the Detection Results Mean?

• PageNumber: Zero-based index of the page.
• RotationAngle: Rotation angle in degrees. Use with the Rotate method to correct orientation.
• HighConfidence: Confidence level in the orientation result for handling edge cases.

When Should I Use High Confidence Values?

The HighConfidence property is crucial for ambiguous or low-quality documents where rotation detection may be uncertain. Documents with sparse text, unusual layouts, or poor scan quality often return lower confidence scores. In these cases, implement additional validation or apply image quality correction filters before detection.

Use this value to implement fallback strategies or manual review for pages with low confidence. For instance, if confidence falls below 80%, process the page with multiple orientations and compare OCR results, or flag for manual review. IronOCR's computer vision features help identify text regions more accurately in challenging documents.

How Do I Correct Detected Rotation?

After identifying the rotation angle, use the Rotate method on your OcrInput object to correct orientation before OCR. This ensures optimal text recognition accuracy. For comprehensive orientation fixes, see the image orientation correction guide. Here's the correction process:

// Apply rotation correction based on detection results
if (result.RotationAngle != 0)
{
    input.Rotate(360 - result.RotationAngle); // Rotate back to 0°
}

// Apply rotation correction based on detection results
if (result.RotationAngle != 0)
{
    input.Rotate(360 - result.RotationAngle); // Rotate back to 0°
}

IRON VB CONVERTER ERROR developers@ironsoftware.com

For documents requiring additional preprocessing, consider the OcrInput Class which provides extensive document preparation methods before OCR processing.

Which Detection Mode Should I Choose?

Four speed options are available for OrientationDetectionMode:

• Fast: High-speed detection with lower accuracy. Ideal for drafts or bulk processing where speed is crucial. Default for DetectPageOrientation. Handles thousands of pages efficiently with multithreading support.

• Balanced: Balanced speed and accuracy. Suitable for production tasks. Uses AdvancedScan extension capabilities for improved accuracy while maintaining performance.

• Detailed: Low speed, high accuracy. Best for precise or critical tasks, especially documents with complex layouts or mixed content.

• ExtremeDetailed: Slowest speed, highest accuracy. Use only when Detailed is insufficient or text is heavily skewed and distorted.

What Are Common Performance Considerations?

Performance varies significantly between modes. Fast mode processes hundreds of pages per minute; ExtremeDetailed may take seconds per page. Choose based on accuracy requirements and time constraints. For optimal performance:

1. Image Resolution: Higher DPI settings improve accuracy but increase processing time. 150-300 DPI typically suffices for rotation detection.

2. Document Type: Text-dense documents process faster and more accurately than sparse layouts. Use the Filter Wizard to optimize image quality before detection.

3. Resource Usage: Monitor memory usage when processing large batches. Implement progress tracking to provide feedback and manage system resources.

4. Parallel Processing: For bulk operations, use IronOCR's multithreading to process multiple documents simultaneously while maintaining accuracy.

How Do I Handle Mixed-Orientation Documents?

For mixed-orientation documents, process each page individually with DetectPageOrientation, then apply page-by-page rotation corrections before OCR. This ensures proper orientation regardless of initial state. Here's an effective approach:

// Process each page with individual rotation detection
for (int i = 0; i < results.Count; i++)
{
    var pageResult = results[i];

    // Apply rotation only to pages that need it
    if (pageResult.RotationAngle != 0 && pageResult.HighConfidence)
    {
        // Correct the specific page
        input.Pages[i].Rotate(360 - pageResult.RotationAngle);
    }
}

// Process each page with individual rotation detection
for (int i = 0; i < results.Count; i++)
{
    var pageResult = results[i];

    // Apply rotation only to pages that need it
    if (pageResult.RotationAngle != 0 && pageResult.HighConfidence)
    {
        // Correct the specific page
        input.Pages[i].Rotate(360 - pageResult.RotationAngle);
    }
}

IRON VB CONVERTER ERROR developers@ironsoftware.com

For complex scenarios involving scanned documents with varying quality or multi-page TIFFs, preprocess each page individually for optimal results.

When processing mixed-format inputs, the OcrResult Class provides detailed page information, enabling sophisticated error handling and quality control workflows. For high-throughput production environments, explore Fast OCR Configuration options to balance speed and accuracy.

If processing documents containing both text and barcodes, use IronOCR's OCR with Barcode & QR Reading capabilities to extract all information in a single pass, improving efficiency.