AWS OCR与Azure OCR（OCR功能比较）

Overview of AWS OCR

AWS Textract is Amazon’s fully managed OCR service designed for text extraction from scanned documents, forms, tables, and more. Integrated deeply within the AWS ecosystem, Textract is optimized for use in large-scale cloud solutions and supports both real-time and batch document processing.

Core Features

• Document text detection and extraction: AWS Textract is highly accurate, particularly in structured documents like forms or tables. It not only extracts raw text but also identifies elements like checkboxes, tables, and key-value pairs.
• Supported file types: AWS Textract supports a variety of image formats such as PNG, JPEG, and TIFF, alongside PDF.
• Table and form data extraction: One of Textract’s most notable features is its ability to accurately identify and extract tabular data and form fields, making it ideal for extracting data at scale.

Performance and Speed

AWS Textract delivers excellent performance, particularly for large-scale batch processing. It can handle extensive datasets efficiently, although real-time processing may exhibit slight delays depending on the document volume.

Integration and API Usability

Textract integrates seamlessly with other AWS services, such as S3, Lambda, and Rekognition, providing a cohesive experience for developers working in the AWS environment. Here's a basic C# example of how you might use Textract with AWS SDK:

// Import necessary AWS Textract and other AWS SDK packages
using Amazon.Textract;
using Amazon.Textract.Model;
using Amazon;

public async Task DetectTextFromDocumentAsync(string bucketName, string documentName)
{
    // Create an Amazon Textract client
    var textractClient = new AmazonTextractClient(RegionEndpoint.USEast1);

    // Prepare the request with the document location in S3
    var request = new DetectDocumentTextRequest
    {
        Document = new Document
        {
            S3Object = new S3Object
            {
                Bucket = bucketName,
                Name = documentName
            }
        }
    };

    // Send request to Textract and await response
    var response = await textractClient.DetectDocumentTextAsync(request);

    // Iterate through the detected blocks of text and print them
    foreach (var block in response.Blocks)
    {
        if (block.BlockType == BlockType.LINE)
        {
            Console.WriteLine($"Detected text: {block.Text}");
        }
    }
}

// Import necessary AWS Textract and other AWS SDK packages
using Amazon.Textract;
using Amazon.Textract.Model;
using Amazon;

public async Task DetectTextFromDocumentAsync(string bucketName, string documentName)
{
    // Create an Amazon Textract client
    var textractClient = new AmazonTextractClient(RegionEndpoint.USEast1);

    // Prepare the request with the document location in S3
    var request = new DetectDocumentTextRequest
    {
        Document = new Document
        {
            S3Object = new S3Object
            {
                Bucket = bucketName,
                Name = documentName
            }
        }
    };

    // Send request to Textract and await response
    var response = await textractClient.DetectDocumentTextAsync(request);

    // Iterate through the detected blocks of text and print them
    foreach (var block in response.Blocks)
    {
        if (block.BlockType == BlockType.LINE)
        {
            Console.WriteLine($"Detected text: {block.Text}");
        }
    }
}

' Import necessary AWS Textract and other AWS SDK packages
Imports Amazon.Textract
Imports Amazon.Textract.Model
Imports Amazon

Public Async Function DetectTextFromDocumentAsync(ByVal bucketName As String, ByVal documentName As String) As Task
	' Create an Amazon Textract client
	Dim textractClient = New AmazonTextractClient(RegionEndpoint.USEast1)

	' Prepare the request with the document location in S3
	Dim request = New DetectDocumentTextRequest With {
		.Document = New Document With {
			.S3Object = New S3Object With {
				.Bucket = bucketName,
				.Name = documentName
			}
		}
	}

	' Send request to Textract and await response
	Dim response = Await textractClient.DetectDocumentTextAsync(request)

	' Iterate through the detected blocks of text and print them
	For Each block In response.Blocks
		If block.BlockType = BlockType.LINE Then
			Console.WriteLine($"Detected text: {block.Text}")
		End If
	Next block
End Function

Pricing

AWS Textract follows a pay-per-use pricing model, where you're billed based on the number of pages processed. Pricing can quickly accumulate for large projects, though it's cost-effective for on-demand usage.

Overview of Azure OCR

Azure Cognitive Services' OCR solution is designed to extract text from images and PDFs and can be integrated into Azure-based applications with ease. It’s suitable for document workflows in cloud and hybrid environments and can be tailored to handle large-scale deployments.

Core Features

• Text extraction accuracy: Azure OCR boasts high accuracy, especially with complex documents like invoices, receipts, and ID cards. It supports over 25 languages, making it ideal for multilingual applications.
• Supported file types: Azure OCR processes images in JPEG, PNG, BMP, PDF, and TIFF formats.
• Multilingual support: Azure OCR can recognize text in many different languages, giving it an advantage when dealing with global projects or applications.
• Form Recognizer: The Azure Form Recognizer is a powerful tool within Azure Cognitive Services that enables developers to extract structured data from forms, invoices, and other documents, enhancing automation and data processing capabilities.

Performance and Speed

Azure OCR excels at real-time processing with an efficient architecture that supports rapid text extraction. Batch processing capabilities are also robust, with Azure’s scalable cloud infrastructure ensuring smooth operations even during peak loads.

Integration and API Usability

Azure OCR integrates tightly with other Azure services like Azure Blob Storage and Azure Functions, making it simple to build end-to-end workflows. The service is accessible through the REST API, and here’s an example in C#:

// Import necessary Azure Cognitive Services packages
using Microsoft.Azure.CognitiveServices.Vision.ComputerVision;
using Microsoft.Azure.CognitiveServices.Vision.ComputerVision.Models;
using System.IO;
using System.Threading.Tasks;

public async Task RecognizeTextInImageAsync(Stream imageStream, string endpoint, string apiKey)
{
    // Create a Computer Vision client
    var client = new ComputerVisionClient(new ApiKeyServiceClientCredentials(apiKey))
    {
        Endpoint = endpoint
    };

    // Call the API with the image stream and read printed text
    var ocrResult = await client.RecognizePrintedTextInStreamAsync(true, imageStream);

    // Iterate over the OCR result regions, lines, and words, printing them
    foreach (var region in ocrResult.Regions)
    {
        foreach (var line in region.Lines)
        {
            foreach (var word in line.Words)
            {
                Console.WriteLine(word.Text);
            }
        }
    }
}

// Import necessary Azure Cognitive Services packages
using Microsoft.Azure.CognitiveServices.Vision.ComputerVision;
using Microsoft.Azure.CognitiveServices.Vision.ComputerVision.Models;
using System.IO;
using System.Threading.Tasks;

public async Task RecognizeTextInImageAsync(Stream imageStream, string endpoint, string apiKey)
{
    // Create a Computer Vision client
    var client = new ComputerVisionClient(new ApiKeyServiceClientCredentials(apiKey))
    {
        Endpoint = endpoint
    };

    // Call the API with the image stream and read printed text
    var ocrResult = await client.RecognizePrintedTextInStreamAsync(true, imageStream);

    // Iterate over the OCR result regions, lines, and words, printing them
    foreach (var region in ocrResult.Regions)
    {
        foreach (var line in region.Lines)
        {
            foreach (var word in line.Words)
            {
                Console.WriteLine(word.Text);
            }
        }
    }
}

' Import necessary Azure Cognitive Services packages
Imports Microsoft.Azure.CognitiveServices.Vision.ComputerVision
Imports Microsoft.Azure.CognitiveServices.Vision.ComputerVision.Models
Imports System.IO
Imports System.Threading.Tasks

Public Async Function RecognizeTextInImageAsync(ByVal imageStream As Stream, ByVal endpoint As String, ByVal apiKey As String) As Task
	' Create a Computer Vision client
	Dim client = New ComputerVisionClient(New ApiKeyServiceClientCredentials(apiKey)) With {.Endpoint = endpoint}

	' Call the API with the image stream and read printed text
	Dim ocrResult = Await client.RecognizePrintedTextInStreamAsync(True, imageStream)

	' Iterate over the OCR result regions, lines, and words, printing them
	For Each region In ocrResult.Regions
		For Each line In region.Lines
			For Each word In line.Words
				Console.WriteLine(word.Text)
			Next word
		Next line
	Next region
End Function

Pricing

Azure OCR offers tiered pricing, based on the number of transactions. It's generally considered cost-effective for enterprises that already leverage Azure infrastructure, though pricing can rise significantly for large datasets.

Overview of IronOCR

IronOCR is a robust third-party OCR tool library designed for .NET developers. It allows for both on-premise and cloud-based implementations, offering more flexibility than AWS or Azure for developers who need tight control over their OCR tools.

Core Features

• Text extraction quality: Extract data from your images and PDF files with ease with this robust tool. IronOCR has high accuracy for print text, excelling in PDF text extraction.
• Image filters: Edit those noisy scanned documents and images with IronOCR's image correction filters, which can denoise images, sharpen them, enhance them, and more!
• File type and language support: IronOCR supports multiple image formats (JPG, GIF, TIFF, BMP) and PDFs, with extensive support for over 100 languages.
• Specialized capabilities: It provides advanced capabilities such as PDF OCR and barcode reading, which are missing in some cloud providers' offerings.

Performance and Speed

IronOCR is optimized for fast text extraction, especially when running on dedicated hardware. For developers needing to process data locally or in hybrid cloud scenarios, IronOCR is an excellent choice, offering high performance even in resource-constrained environments.

Integration and API Usability

IronOCR is highly versatile and easy to use with C#. Here’s a simple example:

// Import IronOcr namespace
using IronOcr;

public class OCRDemo
{
    public void PerformOCR(string imagePath)
    {
        // Create a new instance of IronTesseract
        var ocr = new IronTesseract();

        // Create a new IronOCR image input from the specified image filepath
        using var input = new OcrInput(imagePath);

        // Setting the OCR language (for example, English)
        ocr.Language = OcrLanguage.English;

        // Reads the text from the provided OcrImageInput object and returns an OcrResult object containing the extracted text
        OcrResult result = ocr.Read(input);

        // Writing all of the text to a new text file and saving it
        File.WriteAllText("result.txt", result.Text);
    }
}

// Import IronOcr namespace
using IronOcr;

public class OCRDemo
{
    public void PerformOCR(string imagePath)
    {
        // Create a new instance of IronTesseract
        var ocr = new IronTesseract();

        // Create a new IronOCR image input from the specified image filepath
        using var input = new OcrInput(imagePath);

        // Setting the OCR language (for example, English)
        ocr.Language = OcrLanguage.English;

        // Reads the text from the provided OcrImageInput object and returns an OcrResult object containing the extracted text
        OcrResult result = ocr.Read(input);

        // Writing all of the text to a new text file and saving it
        File.WriteAllText("result.txt", result.Text);
    }
}

' Import IronOcr namespace
Imports IronOcr

Public Class OCRDemo
	Public Sub PerformOCR(ByVal imagePath As String)
		' Create a new instance of IronTesseract
		Dim ocr = New IronTesseract()

		' Create a new IronOCR image input from the specified image filepath
		Dim input = New OcrInput(imagePath)

		' Setting the OCR language (for example, English)
		ocr.Language = OcrLanguage.English

		' Reads the text from the provided OcrImageInput object and returns an OcrResult object containing the extracted text
		Dim result As OcrResult = ocr.Read(input)

		' Writing all of the text to a new text file and saving it
		File.WriteAllText("result.txt", result.Text)
	End Sub
End Class

Pricing

IronOCR’s licensing model is more flexible than AWS or Azure. You pay a one-time fee for a perpetual license, which can be more cost-effective for small to medium-sized projects. As a bonus, IronOCR offers a free trial, with options available for enterprises.