How to Define a Specific OCR Region of an Image in C#

How Do I Perform OCR on a Specific Region?

To define a specific OCR region, you create a Rectangle object from the IronSoftware.Drawing namespace. This object requires four values: the x-coordinate, the y-coordinate, the width, and the height, all in pixels. The (x, y) coordinates represent the top-left corner of your desired area.

When you load your image using LoadImage, you pass this Rectangle as the second parameter. IronOCR will then restrict its OCR process to only the pixels within that bounding box.

Regional OCR is particularly useful when working with structured documents like invoices, scanned forms, or identity documents where specific information always appears in predictable locations. By limiting the OCR to just the relevant areas, you can significantly improve processing speed and reduce false positives from unrelated text.

To find the coordinates for your Rectangle, you can use a simple image editor like MS Paint. Open your input image, hover your mouse over the top-left and bottom-right corners of the specified region, and note the (x, y) pixel coordinates. You can then calculate the rectangle's properties: (x1, y1, width, height), where width = x2-x1 and height = y2-y1.

What Image Should I Use for Testing?

We'll use a sample image with three paragraphs. Our goal is to extract only the second paragraph and ignore the rest of the text. This demonstrates a common scenario where you need to extract specific fields or sections from a larger document.

:path=/static-assets/ocr/content-code-examples/how-to/ocr-region-of-an-image.cs

using IronOcr;
using IronSoftware.Drawing;
using System;

var ocrTesseract = new IronTesseract();
using var ocrInput = new OcrInput();

// Define the specific region as a Rectangle
// (x, y) is the top-left corner.
var ContentArea = new Rectangle(x: 215, y: 1250, width: 1335, height: 280);

ocrInput.LoadImage("region-input.png", ContentArea);

var ocrResult = ocrTesseract.Read(ocrInput);

// Print the extracted text
Console.WriteLine(ocrResult.Text);

using IronOcr;
using IronSoftware.Drawing;

var ocr = new IronTesseract();
using var input = new OcrInput();

// Define multiple regions for different form fields
var nameField = new Rectangle(x: 100, y: 200, width: 300, height: 50);
var dateField = new Rectangle(x: 100, y: 300, width: 200, height: 50);
var amountField = new Rectangle(x: 400, y: 500, width: 150, height: 50);

// Load the same image multiple times with different regions
input.LoadImage("form.png", nameField);
var nameResult = ocr.Read(input);

input.Clear();
input.LoadImage("form.png", dateField);
var dateResult = ocr.Read(input);

input.Clear();
input.LoadImage("form.png", amountField);
var amountResult = ocr.Read(input);

// Process each field separately
Console.WriteLine($"Name: {nameResult.Text}");
Console.WriteLine($"Date: {dateResult.Text}");
Console.WriteLine($"Amount: {amountResult.Text}");

using IronOcr;
using IronSoftware.Drawing;

var ocr = new IronTesseract();
using var input = new OcrInput();

// Define multiple regions for different form fields
var nameField = new Rectangle(x: 100, y: 200, width: 300, height: 50);
var dateField = new Rectangle(x: 100, y: 300, width: 200, height: 50);
var amountField = new Rectangle(x: 400, y: 500, width: 150, height: 50);

// Load the same image multiple times with different regions
input.LoadImage("form.png", nameField);
var nameResult = ocr.Read(input);

input.Clear();
input.LoadImage("form.png", dateField);
var dateResult = ocr.Read(input);

input.Clear();
input.LoadImage("form.png", amountField);
var amountResult = ocr.Read(input);

// Process each field separately
Console.WriteLine($"Name: {nameResult.Text}");
Console.WriteLine($"Date: {dateResult.Text}");
Console.WriteLine($"Amount: {amountResult.Text}");

What Results Can I Expect?

As you can see from the console output, only the second paragraph is processed by the OCR. This targeted approach ensures that irrelevant text from other parts of the image doesn't interfere with your results.

The accuracy of regional OCR depends on several factors:

• Image quality: Higher resolution images generally yield better results. Consider using DPI settings to optimize your images.
• Text orientation: Ensure text is properly aligned. Use page rotation detection if needed.
• Contrast and clarity: Apply image correction filters to enhance text readability.

How Can I Verify My Coordinates Are Correct?

To ensure you've selected the correct coordinates for the input image, you can visualize the ContentArea you defined. A simple way to do this is to draw the rectangle on the input image and save it as a new file with StampCropRectangleAndSaveAs. This helps you debug and fine-tune the coordinates for optimal performance.

This visualization technique is especially helpful when working with complex layouts or when you need to highlight specific text areas for quality assurance purposes.

Here is the output image after drawing the specified bounding box on our example input image from above.

How Do I Visualize the Selected Region?

:path=/static-assets/ocr/content-code-examples/how-to/ocr-region-of-an-image-highlighted.cs

using IronOcr;
using IronSoftware.Drawing;

var ocrTesseract = new IronTesseract();
using var ocrInput = new OcrInput();

// Define the specific rectangular area to scan within the image.
// The coordinates are in pixels: (x, y) is the top-left corner of the rectangle.
var ContentArea = new Rectangle(x: 4, y: 59, width: 365, height: 26);

ocrInput.LoadImage("region-input.png", ContentArea);

var ocrResult = ocrTesseract.Read(ocrInput);

// Draws the rectangle from above in a blue bounding box on the image for visualization.
ocrInput.StampCropRectangleAndSaveAs(ContentArea, Color.Aqua, "region-input.png");

$vbLabelText   $csharpLabel

What Does the Visualization Look Like?

The light blue rectangle confirms that we have correctly isolated the second paragraph for processing.

When Should I Use Regional OCR?

Regional OCR is ideal for several common scenarios:

1. Form Processing: When extracting specific fields from standardized forms where data appears in consistent locations.
2. Invoice Processing: To extract specific values like totals, dates, or invoice numbers without processing the entire document.
3. License Plates: When using license plate recognition, focus only on the plate area.
4. Identity Documents: Extract specific fields from passports or ID cards.
5. Screenshots: When capturing text from specific UI elements in screenshots.

Best Practices for Regional OCR

To achieve optimal results with regional OCR:

1. Add padding: Include a small buffer around your text to ensure no characters are cut off at the edges.
2. Test with sample images: Always verify your coordinates with representative samples before processing large batches.
3. Handle variations: Account for slight positioning variations in scanned documents by making your regions slightly larger than necessary.
4. Optimize performance: For multithreaded processing, process different regions in parallel.
5. Monitor confidence: Check the result confidence scores to ensure accuracy.

By focusing OCR processing on specific regions, you can significantly improve both the speed and accuracy of your text extraction tasks. This targeted approach is essential for building efficient document processing workflows in your .NET applications.