OCR API Microsoft Azure Vision 与IronOCR：哪个更擅长处理文档图像？

// Tesseract requires every one of these steps to be written manually
public static string ExtractWithPreprocessing(string imagePath)
{
    using (var original = new Bitmap(imagePath))
    {
        // Step 2: Convert to grayscale (~25 lines)
        using (var grayscale = ConvertToGrayscale(original))
        {
            // Step 3: Apply contrast enhancement (~15 lines)
            using (var enhanced = EnhanceContrast(grayscale))
            {
                // Step 4: Binarize — convert to black and white (~15 lines)
                using (var binarized = Binarize(enhanced, 128))
                {
                    // Step 5: Remove noise (~25 lines)
                    using (var denoised = RemoveNoise(binarized))
                    {
                        // Step 6: 德斯丘 if rotated (~50 lines, simplified)
                        using (var deskewed = Deskew(denoised))
                        {
                            // Step 7: Scale to 300 DPI (~20 lines)
                            using (var scaled = ScaleToDpi(deskewed, 300))
                            {
                                return RunTesseract(scaled);  // Save to temp file, load Pix, process
                            }
                        }
                    }
                }
            }
        }
    }
}

Imports System.Drawing

Public Class ImageProcessor
    Public Shared Function ExtractWithPreprocessing(imagePath As String) As String
        Using original As New Bitmap(imagePath)
            ' Step 2: Convert to grayscale (~25 lines)
            Using grayscale As Bitmap = ConvertToGrayscale(original)
                ' Step 3: Apply contrast enhancement (~15 lines)
                Using enhanced As Bitmap = EnhanceContrast(grayscale)
                    ' Step 4: Binarize — convert to black and white (~15 lines)
                    Using binarized As Bitmap = Binarize(enhanced, 128)
                        ' Step 5: Remove noise (~25 lines)
                        Using denoised As Bitmap = RemoveNoise(binarized)
                            ' Step 6: Deskew if rotated (~50 lines, simplified)
                            Using deskewed As Bitmap = Deskew(denoised)
                                ' Step 7: Scale to 300 DPI (~20 lines)
                                Using scaled As Bitmap = ScaleToDpi(deskewed, 300)
                                    Return RunTesseract(scaled) ' Save to temp file, load Pix, process
                                End Using
                            End Using
                        End Using
                    End Using
                End Using
            End Using
        End Using
    End Function

    ' Placeholder methods for image processing steps
    Private Shared Function ConvertToGrayscale(original As Bitmap) As Bitmap
        ' Implementation here
        Return original
    End Function

    Private Shared Function EnhanceContrast(grayscale As Bitmap) As Bitmap
        ' Implementation here
        Return grayscale
    End Function

    Private Shared Function Binarize(enhanced As Bitmap, threshold As Integer) As Bitmap
        ' Implementation here
        Return enhanced
    End Function

    Private Shared Function RemoveNoise(binarized As Bitmap) As Bitmap
        ' Implementation here
        Return binarized
    End Function

    Private Shared Function Deskew(denoised As Bitmap) As Bitmap
        ' Implementation here
        Return denoised
    End Function

    Private Shared Function ScaleToDpi(deskewed As Bitmap, dpi As Integer) As Bitmap
        ' Implementation here
        Return deskewed
    End Function

    Private Shared Function RunTesseract(scaled As Bitmap) As String
        ' Implementation here
        Return String.Empty
    End Function
End Class

// image-preprocessing-tesseract.cs — the actual implementation pattern
private static Bitmap ConvertToGrayscale(Bitmap original)
{
    var result = new Bitmap(original.Width, original.Height);
    using (var graphics = Graphics.FromImage(result))
    {
        var colorMatrix = new ColorMatrix(new float[][]
        {
            new float[] { 0.299f, 0.299f, 0.299f, 0, 0 },
            new float[] { 0.587f, 0.587f, 0.587f, 0, 0 },
            new float[] { 0.114f, 0.114f, 0.114f, 0, 0 },
            new float[] { 0, 0, 0, 1, 0 },
            new float[] { 0, 0, 0, 0, 1 }
        });
        using (var attributes = new ImageAttributes())
        {
            attributes.SetColorMatrix(colorMatrix);
            graphics.DrawImage(original,
                new Rectangle(0, 0, original.Width, original.Height),
                0, 0, original.Width, original.Height,
                GraphicsUnit.Pixel, attributes);
        }
    }
    return result;
}

private static Bitmap EnhanceContrast(Bitmap image)
{
    var result = new Bitmap(image.Width, image.Height);
    float contrast = 1.5f;
    for (int y = 0; y < image.Height; y++)
    {
        for (int x = 0; x < image.Width; x++)
        {
            var pixel = image.GetPixel(x, y);
            int r = Clamp((int)((pixel.R - 128) * contrast + 128));
            int g = Clamp((int)((pixel.G - 128) * contrast + 128));
            int b = Clamp((int)((pixel.B - 128) * contrast + 128));
            result.SetPixel(x, y, Color.FromArgb(r, g, b));
        }
    }
    return result;
}

private static Bitmap RemoveNoise(Bitmap image)
{
    var result = new Bitmap(image.Width, image.Height);
    int kernelSize = 3;
    int radius = kernelSize / 2;
    for (int y = radius; y < image.Height - radius; y++)
    {
        for (int x = radius; x < image.Width - radius; x++)
        {
            var pixels = new List<int>();
            for (int ky = -radius; ky <= radius; ky++)
                for (int kx = -radius; kx <= radius; kx++)
                    pixels.Add(image.GetPixel(x + kx, y + ky).R);
            pixels.Sort();
            int median = pixels[pixels.Count / 2];
            result.SetPixel(x, y, Color.FromArgb(median, median, median));
        }
    }
    return result;
}

// After all preprocessing, save to temp file — Tesseract requires a file path
private static string RunTesseract(Bitmap preprocessed)
{
    string tempPath = Path.GetTempFileName() + ".png";
    try
    {
        preprocessed.Save(tempPath, ImageFormat.Png);
        using (var engine = new TesseractEngine(TessDataPath, "eng", EngineMode.Default))
        using (var img = Pix.LoadFromFile(tempPath))
        using (var page = engine.Process(img))
            return page.GetText();
    }
    finally
    {
        if (File.Exists(tempPath)) File.Delete(tempPath);
    }
}

Imports System.Drawing
Imports System.Drawing.Imaging
Imports Tesseract
Imports System.IO

' image-preprocessing-tesseract.vb — the actual implementation pattern
Private Shared Function ConvertToGrayscale(original As Bitmap) As Bitmap
    Dim result As New Bitmap(original.Width, original.Height)
    Using graphics As Graphics = Graphics.FromImage(result)
        Dim colorMatrix As New ColorMatrix(New Single()() {
            New Single() {0.299F, 0.299F, 0.299F, 0, 0},
            New Single() {0.587F, 0.587F, 0.587F, 0, 0},
            New Single() {0.114F, 0.114F, 0.114F, 0, 0},
            New Single() {0, 0, 0, 1, 0},
            New Single() {0, 0, 0, 0, 1}
        })
        Using attributes As New ImageAttributes()
            attributes.SetColorMatrix(colorMatrix)
            graphics.DrawImage(original, New Rectangle(0, 0, original.Width, original.Height), 0, 0, original.Width, original.Height, GraphicsUnit.Pixel, attributes)
        End Using
    End Using
    Return result
End Function

Private Shared Function EnhanceContrast(image As Bitmap) As Bitmap
    Dim result As New Bitmap(image.Width, image.Height)
    Dim contrast As Single = 1.5F
    For y As Integer = 0 To image.Height - 1
        For x As Integer = 0 To image.Width - 1
            Dim pixel As Color = image.GetPixel(x, y)
            Dim r As Integer = Clamp(CInt((pixel.R - 128) * contrast + 128))
            Dim g As Integer = Clamp(CInt((pixel.G - 128) * contrast + 128))
            Dim b As Integer = Clamp(CInt((pixel.B - 128) * contrast + 128))
            result.SetPixel(x, y, Color.FromArgb(r, g, b))
        Next
    Next
    Return result
End Function

Private Shared Function RemoveNoise(image As Bitmap) As Bitmap
    Dim result As New Bitmap(image.Width, image.Height)
    Dim kernelSize As Integer = 3
    Dim radius As Integer = kernelSize \ 2
    For y As Integer = radius To image.Height - radius - 1
        For x As Integer = radius To image.Width - radius - 1
            Dim pixels As New List(Of Integer)()
            For ky As Integer = -radius To radius
                For kx As Integer = -radius To radius
                    pixels.Add(image.GetPixel(x + kx, y + ky).R)
                Next
            Next
            pixels.Sort()
            Dim median As Integer = pixels(pixels.Count \ 2)
            result.SetPixel(x, y, Color.FromArgb(median, median, median))
        Next
    Next
    Return result
End Function

' After all preprocessing, save to temp file — Tesseract requires a file path
Private Shared Function RunTesseract(preprocessed As Bitmap) As String
    Dim tempPath As String = Path.GetTempFileName() & ".png"
    Try
        preprocessed.Save(tempPath, ImageFormat.Png)
        Using engine As New TesseractEngine(TessDataPath, "eng", EngineMode.Default)
            Using img As Pix = Pix.LoadFromFile(tempPath)
                Using page As Page = engine.Process(img)
                    Return page.GetText()
                End Using
            End Using
        End Using
    Finally
        If File.Exists(tempPath) Then File.Delete(tempPath)
    End Try
End Function

Private Shared Function Clamp(value As Integer) As Integer
    Return Math.Max(0, Math.Min(255, value))
End Function

Private Const TessDataPath As String = "path_to_tessdata" ' Define the path to tessdata directory

// dotnet add package IronOcr
using IronOcr;

using var input = new OcrInput();
input.LoadImage("low-quality-scan.jpg");

input.Deskew();           // Detects and corrects skew angle automatically
input.DeNoise();          // Removes scanner artifacts and specks
input.Contrast();         // Enhances contrast for character separation
input.Binarize();         // Converts to black and white with adaptive threshold
input.EnhanceResolution(300);  // Scales to 300 DPI for optimal recognition

var result = new IronTesseract().Read(input);
Console.WriteLine($"Confidence: {result.Confidence}%");
Console.WriteLine(result.Text);

Imports IronOcr

Dim input As New OcrInput()
input.LoadImage("low-quality-scan.jpg")

input.Deskew()           ' Detects and corrects skew angle automatically
input.DeNoise()          ' Removes scanner artifacts and specks
input.Contrast()         ' Enhances contrast for character separation
input.Binarize()         ' Converts to black and white with adaptive threshold
input.EnhanceResolution(300)  ' Scales to 300 DPI for optimal recognition

Using input
    Dim result = New IronTesseract().Read(input)
    Console.WriteLine($"Confidence: {result.Confidence}%")
    Console.WriteLine(result.Text)
End Using

// Automatic preprocessing applied internally — no explicit filter calls needed
var text = new IronTesseract().Read("scanned-invoice.jpg").Text;

Imports IronTesseract

' Automatic preprocessing applied internally — no explicit filter calls needed
Dim text As String = New IronTesseract().Read("scanned-invoice.jpg").Text

// Tesseract PDF processing — from pdf-ocr-processing-tesseract.cs
// Requires: PdfiumViewer NuGet + pdfium native DLL deployed to application directory
// NuGet: PdfiumViewer, PdfiumViewer.Native.x64

using PdfiumViewer;

public static string ExtractFromPdfWithPdfium(string pdfPath)
{
    var results = new List<string>();

    using (var document = PdfDocument.Load(pdfPath))
    {
        using (var engine = new TesseractEngine(TessDataPath, "eng", EngineMode.Default))
        {
            for (int pageIndex = 0; pageIndex < document.PageCount; pageIndex++)
            {
                // Render page to image at 300 DPI
                using (var pageImage = document.Render(pageIndex, 300, 300,
                    PdfRenderFlags.CorrectFromDpi))
                {
                    // Tesseract requires a file path — must write to disk first
                    string tempPath = Path.GetTempFileName() + ".png";
                    try
                    {
                        pageImage.Save(tempPath);
                        using (var img = Pix.LoadFromFile(tempPath))
                        using (var page = engine.Process(img))
                            results.Add(page.GetText());
                    }
                    finally
                    {
                        File.Delete(tempPath);  // Must clean up or disk fills
                    }
                }
            }
        }
    }

    return string.Join("\n\n--- Page Break ---\n\n", results);
}

Imports PdfiumViewer
Imports Tesseract

Public Shared Function ExtractFromPdfWithPdfium(pdfPath As String) As String
    Dim results As New List(Of String)()

    Using document = PdfDocument.Load(pdfPath)
        Using engine = New TesseractEngine(TessDataPath, "eng", EngineMode.Default)
            For pageIndex As Integer = 0 To document.PageCount - 1
                ' Render page to image at 300 DPI
                Using pageImage = document.Render(pageIndex, 300, 300, PdfRenderFlags.CorrectFromDpi)
                    ' Tesseract requires a file path — must write to disk first
                    Dim tempPath As String = Path.GetTempFileName() & ".png"
                    Try
                        pageImage.Save(tempPath)
                        Using img = Pix.LoadFromFile(tempPath)
                            Using page = engine.Process(img)
                                results.Add(page.GetText())
                            End Using
                        End Using
                    Finally
                        File.Delete(tempPath)  ' Must clean up or disk fills
                    End Try
                End Using
            Next
        End Using
    End Using

    Return String.Join(vbCrLf & vbCrLf & "--- Page Break ---" & vbCrLf & vbCrLf, results)
End Function

// dotnet add package IronOcr
using IronOcr;

// Scanned PDF — direct load, no rendering library required
var result = new IronTesseract().Read("scanned-contract.pdf");
Console.WriteLine(result.Text);

// Password-protected PDF — one additional parameter
using var input = new OcrInput();
input.LoadPdf("encrypted.pdf", Password: "secret");
var protectedResult = new IronTesseract().Read(input);

// Specific page range from a 200-page document
using var rangeInput = new OcrInput();
rangeInput.LoadPdfPages("large-report.pdf", 1, 10);
var rangeResult = new IronTesseract().Read(rangeInput);

// Create searchable PDF with embedded text layer
var searchable = new IronTesseract().Read("scanned-invoice.pdf");
searchable.SaveAsSearchablePdf("searchable-invoice.pdf");

Imports IronOcr

' Scanned PDF — direct load, no rendering library required
Dim result = New IronTesseract().Read("scanned-contract.pdf")
Console.WriteLine(result.Text)

' Password-protected PDF — one additional parameter
Using input As New OcrInput()
    input.LoadPdf("encrypted.pdf", Password:="secret")
    Dim protectedResult = New IronTesseract().Read(input)
End Using

' Specific page range from a 200-page document
Using rangeInput As New OcrInput()
    rangeInput.LoadPdfPages("large-report.pdf", 1, 10)
    Dim rangeResult = New IronTesseract().Read(rangeInput)
End Using

' Create searchable PDF with embedded text layer
Dim searchable = New IronTesseract().Read("scanned-invoice.pdf")
searchable.SaveAsSearchablePdf("searchable-invoice.pdf")

// Must exist before initialization:
// ./tessdata/eng.traineddata   (~15 MB)
// ./tessdata/fra.traineddata   (~15 MB)
// ./tessdata/deu.traineddata   (~15 MB)
// ./tessdata/chi_sim.traineddata  (~45 MB)
// ./tessdata/jpn.traineddata   (~40 MB)
// 10 languages = 200-300 MB to download and manage

public string SafeMultiLanguageOcr(string imagePath, string[] languages)
{
    // Check presence before attempting — runtime failures are worse
    foreach (var lang in languages)
    {
        if (!File.Exists(Path.Combine(TessDataPath, $"{lang}.traineddata")))
        {
            throw new FileNotFoundException(
                $"Missing {lang}.traineddata in {TessDataPath}. " +
                "Download from https://github.com/tesseract-ocr/tessdata");
        }
    }

    var langString = string.Join("+", languages);  // e.g., "eng+fra+deu"
    using var engine = new TesseractEngine(TessDataPath, langString, EngineMode.Default);
    using var img = Pix.LoadFromFile(imagePath);
    using var page = engine.Process(img);
    return page.GetText();
}

Imports System.IO
Imports Tesseract

Public Function SafeMultiLanguageOcr(imagePath As String, languages As String()) As String
    ' Check presence before attempting — runtime failures are worse
    For Each lang In languages
        If Not File.Exists(Path.Combine(TessDataPath, $"{lang}.traineddata")) Then
            Throw New FileNotFoundException(
                $"Missing {lang}.traineddata in {TessDataPath}. " &
                "Download from https://github.com/tesseract-ocr/tessdata")
        End If
    Next

    Dim langString = String.Join("+", languages)  ' e.g., "eng+fra+deu"
    Using engine As New TesseractEngine(TessDataPath, langString, EngineMode.Default)
        Using img As Pix = Pix.LoadFromFile(imagePath)
            Using page As Page = engine.Process(img)
                Return page.GetText()
            End Using
        End Using
    End Using
End Function

// Install once: dotnet add package IronOcr.Languages.French
// Install once: dotnet add package IronOcr.Languages.German
using IronOcr;

var ocr = new IronTesseract();
ocr.Language = OcrLanguage.French;
ocr.AddSecondaryLanguage(OcrLanguage.German);

var result = ocr.Read("multilingual-document.jpg");

Imports IronOcr

' Install once: dotnet add package IronOcr.Languages.French
' Install once: dotnet add package IronOcr.Languages.German

Dim ocr As New IronTesseract()
ocr.Language = OcrLanguage.French
ocr.AddSecondaryLanguage(OcrLanguage.German)

Dim result = ocr.Read("multilingual-document.jpg")