将 Aspose.OCR 与 Amazon S3 集成,使开发人员能够有效地利用云存储,以便可视性字符识别(OCR)的结果。 这种整合不仅简化了 OCR 数据的管理,而且还提高了可扩展性和可访问性。 在本教程中,我们将通过设置 ASPOSE.OKR 的过程,与 AWS S 3 无缝工作,提供详细的示例和最佳实践。

完整的例子

原則

  • 安装了 .NET 8 (或 .Net 6+) SDK。
  • AWS 帐户与 Amazon S3 访问。
  • 一个布克(例如, my-ocr-demo-bucket)在您最喜欢的区域(下面的例子使用 ap-south-1).
  • (可选) 如果您想要超越评估运行,请删除许可文件。

步骤 1: 为 .NET 设置 Aspose.OCR

安装一个新的控制台应用程序,并添加 NuGet 包。

# Create project
dotnet new console -n OcrS3Demo -f net8.0
cd OcrS3Demo

# Add Aspose.OCR (CPU) OR Aspose.OCR-GPU (pick exactly one)
dotnet add package Aspose.OCR
# dotnet add package Aspose.OCR-GPU   # if you prefer GPU build

# Add AWS S3 SDK
dotnet add package AWSSDK.S3

阿斯波斯提供两 Aspose.OCR (CPU)和 Aspose.OCR-GPU 通过 NuGet 包装;你只需要一个。Aspose 文档)

步骤2:为 .NET 设置 AWS SDK

设置您的 AWS 个人资料并创建一个漏洞(如果您已经拥有一个)。

# Configure credentials (creates ~/.aws/credentials and config)
aws configure --profile ocr-s3
# AWS Access Key ID: AKIA****************
# AWS Secret Access Key: ************************
# Default region name: ap-south-1
# Default output format: json

# Create a bucket in that region (bucket name must be globally unique)
aws s3api create-bucket \
  --bucket my-ocr-demo-bucket \
  --region ap-south-1 \
  --create-bucket-configuration LocationConstraint=ap-south-1

推荐的最低 IAM 政策(加入您的用户/角色)为此教程:

{
  "Version": "2012-10-17",
  "Statement": [
    { "Effect": "Allow", "Action": ["s3:ListBucket"], "Resource": "arn:aws:s3:::my-ocr-demo-bucket" },
    { "Effect": "Allow", "Action": ["s3:GetObject", "s3:PutObject"], "Resource": "arn:aws:s3:::my-ocr-demo-bucket/*" }
  ]
}

AWS SDK for .NET 使用默认认链;设置 AWS_PROFILE=ocr-s3 Core S3 模式(创建、上传、下载)在 AWS 官方 .NET 示例中文档。AWS 文档)

步骤3:启动 Aspose.OCR API

创建一个基本 Program.cs 我们还将设置英语作为语言和文档布局检测(下面显示的所有类型来自当前的Aspose.OCR API表面)。参考.aspose.com)

using System;
using System.IO;
using System.Text;
using System.Threading.Tasks;
using Amazon;
using Amazon.S3;
using Amazon.S3.Model;
using Aspose.OCR;

class Program
{
    static async Task Main(string[] args)
    {
        // Optional: load license if you have one
        // new License().SetLicense("Aspose.Total.lic");

        var ocr = new AsposeOcr();

        var settings = new RecognitionSettings
        {
            // pick your language(s); can combine if needed
            Language = Language.Eng,
            DetectAreasMode = DetectAreasMode.DOCUMENT
        };

        // We'll fill in S3 + OCR steps next...
    }
}

关键 API 我们将使用下一个:

  • AsposeOcr.RecognizeImage(MemoryStream, RecognitionSettings) 返回 A RecognitionResult.
  • RecognitionResult.RecognitionText / GetJson(bool) / Save(...) 請將結果輸出到 TXT/JSON/PDF/DOCX。参考.aspose.com)

步骤4:将图像上传到S3

您可以从磁盘上传图像到 S3 PutObjectAsync(您还可以上传流;两者都由 AWS SDK 支持。

// Configure S3 client (uses your AWS_PROFILE locally)
var region = RegionEndpoint.APSouth1; // change if needed
using var s3 = new AmazonS3Client(region);

// Local image you want to OCR:
string localImagePath = @"D:\samples\invoices\invoice-001.png";
string bucket = "my-ocr-demo-bucket";
string objectKey = "input/invoice-001.png";

// Upload the image to S3
await s3.PutObjectAsync(new PutObjectRequest
{
    BucketName = bucket,
    Key = objectKey,
    FilePath = localImagePath,
    ContentType = "image/png",
    // Optional: enable server-side encryption
    // ServerSideEncryptionMethod = ServerSideEncryptionMethod.AES256
});
Console.WriteLine($"Uploaded {objectKey} to s3://{bucket}.");

查看 AWS 的 .NET S3 上传模式的示例。AWS 文档)

步骤5:在上传图像上完成OCR

将 S3 对象直接流入内存并通过 MemoryStream 到 Aspose.OCR。

// Download S3 object and OCR in-memory (no temp files)
var get = await s3.GetObjectAsync(bucket, objectKey);
await using var s3Stream = get.ResponseStream;
using var ms = new MemoryStream();
await s3Stream.CopyToAsync(ms);
ms.Position = 0;

// Run OCR (with settings → structured result)
RecognitionResult result = ocr.RecognizeImage(ms, settings);

// Or: if you just need plain text and defaults
// string textFast = ocr.RecognizeImage(ms);

string recognizedText = result.RecognitionText;
Console.WriteLine("=== OCR TEXT ===");
Console.WriteLine(recognizedText);

是的 RecognizeImage 加载和 RecognitionResult.RecognitionText 是目前的 API 参考的一部分。参考.aspose.com)

步骤6:在 S3 中存储 OCR 结果

您可以上传平板文本,JSON,甚至由Aspose.OCR制作的PDF/DOCX。

6.a) 保存为清晰的文本

var textKey = "output/invoice-001.txt";
var textBytes = Encoding.UTF8.GetBytes(recognizedText);
await s3.PutObjectAsync(new PutObjectRequest
{
    BucketName = bucket,
    Key = textKey,
    InputStream = new MemoryStream(textBytes),
    ContentType = "text/plain"
});
Console.WriteLine($"Saved OCR text to s3://{bucket}/{textKey}");

6.b) 保存详细的 JSON

var json = result.GetJson(true); // include additional data
var jsonKey = "output/invoice-001.json";
await s3.PutObjectAsync(new PutObjectRequest
{
    BucketName = bucket,
    Key = jsonKey,
    InputStream = new MemoryStream(Encoding.UTF8.GetBytes(json)),
    ContentType = "application/json"
});
Console.WriteLine($"Saved OCR JSON to s3://{bucket}/{jsonKey}");

6.c)保存可搜索的PDF(或DOCX)并将其插入S3

// Export to PDF in-memory, then upload
using var outPdf = new MemoryStream();
result.Save(outPdf, SaveFormat.Pdf, "Arial", PdfOptimizationMode.Basic);
outPdf.Position = 0;

var pdfKey = "output/invoice-001.pdf";
await s3.PutObjectAsync(new PutObjectRequest
{
    BucketName = bucket,
    Key = pdfKey,
    InputStream = outPdf,
    ContentType = "application/pdf"
});
Console.WriteLine($"Saved OCR PDF to s3://{bucket}/{pdfKey}");

出口和储蓄方法(RecognitionResult.Save)和格式(TXT/PDF/DOCX)在官方API参考中。

可选:终端 Program.cs

这里是一个微妙的终端版本,你可以下载到 Program.cs (组合步骤3至6):

using System;
using System.IO;
using System.Text;
using System.Threading.Tasks;
using Amazon;
using Amazon.S3;
using Amazon.S3.Model;
using Aspose.OCR;

class Program
{
    static async Task Main()
    {
        // new License().SetLicense("Aspose.Total.lic"); // optional

        string bucket = "my-ocr-demo-bucket";
        string regionSystemName = "ap-south-1";
        string localImagePath = @"D:\samples\invoices\invoice-001.png";
        string imageKey = "input/invoice-001.png";

        var ocr = new AsposeOcr();
        var settings = new RecognitionSettings
        {
            Language = Language.Eng,
            DetectAreasMode = DetectAreasMode.DOCUMENT
        };

        using var s3 = new AmazonS3Client(RegionEndpoint.GetBySystemName(regionSystemName));

        // Upload original
        await s3.PutObjectAsync(new PutObjectRequest
        {
            BucketName = bucket,
            Key = imageKey,
            FilePath = localImagePath,
            ContentType = "image/png"
        });

        // Get image as stream
        var get = await s3.GetObjectAsync(bucket, imageKey);
        await using var s3Stream = get.ResponseStream;
        using var ms = new MemoryStream();
        await s3Stream.CopyToAsync(ms);
        ms.Position = 0;

        // OCR
        RecognitionResult result = ocr.RecognizeImage(ms, settings);
        string text = result.RecognitionText;

        // Upload text
        await s3.PutObjectAsync(new PutObjectRequest
        {
            BucketName = bucket,
            Key = "output/invoice-001.txt",
            InputStream = new MemoryStream(Encoding.UTF8.GetBytes(text)),
            ContentType = "text/plain"
        });

        // Upload JSON
        string json = result.GetJson(true);
        await s3.PutObjectAsync(new PutObjectRequest
        {
            BucketName = bucket,
            Key = "output/invoice-001.json",
            InputStream = new MemoryStream(Encoding.UTF8.GetBytes(json)),
            ContentType = "application/json"
        });

        // Upload PDF
        using var outPdf = new MemoryStream();
        result.Save(outPdf, SaveFormat.Pdf, "Arial", PdfOptimizationMode.Basic);
        outPdf.Position = 0;
        await s3.PutObjectAsync(new PutObjectRequest
        {
            BucketName = bucket,
            Key = "output/invoice-001.pdf",
            InputStream = outPdf,
            ContentType = "application/pdf"
        });

        Console.WriteLine("OCR complete and results stored in S3.");
    }
}

最佳实践

将 Aspose.OCR 与 AWS S3 集成,提供了许多好处,包括提高数据管理和提高规模性。

  • “安全”

  • 永远不要使用硬代码的秘密 aws configure + AWS_PROFILE 在本地;使用IAM在生产中的角色。

  • 考虑 S3 服务器侧加密(AES256 或 KMS)在结果对象上,并以最少的特权(如上所示)。AWS 文档)

  • 性能

  • 使用 GPU 包(Aspose.OCR-GPU)在 CUDA 可加速 OCR 的硬件上;同一代码,更快的执行。Aspose 文档)

  • 预处理图像为质量(Deskw、Denoise)使用 RecognitionSettings 如果需要的话,先选择正确的。 DetectAreasMode 对于文件. API 选项在参考中显示。参考.aspose.com)

    • 可扩展*

  • 使用 S3 Prefix 如 input/ 和 output/ 按工作,并为可追踪提供 OCR 文件(TXT/JSON/PDF)存储。

  • 允许 S3 版本,如果您想要可审查的历史和旋转。

  • 考虑运行此流量在容器或无服务器(例如,AWS Batch/ECS/Lambda 与 EFS)为平行 OCR 规模。

通过遵循这些指南,您可以有效地将 Aspose.OCR 与 AWS S3 集成,以便简化您的 OCR 工作流,并提高您的应用程序的整体性能。

参考

  • 安东尼亚(S&D) 包和安装选项(Aspose.OCR, Aspose.OCR-GPU). (Aspose 文档)
  • AsposeOcr.RecognizeImage(...) 超重; RecognitionResult.RecognitionText, GetJson, Save(...). (参考.aspose.com)
  • AWS SDK for .NET: S3 创建/上传/下载示例。AWS 文档)

如果你愿意,我也可以添加一个小 Makefile 或 PowerShell 脚本运行此端到端,加上一个 CI 剪辑(GitHub 行动)将结果推到 S3 在承诺。

[4]: https://reference.aspose.com/ocr/net/aspose.ocr/recognitionresult/ “认可结果 ”

More in this category