LlosLab Software Development BlogEngineering notes for PDF, Delphi, and document software developers.

Technical Article

Extracting Images from PDF Documents with PDFium VCL in Delphi

· PDF Programming

PDF documents often contain valuable images—photographs, diagrams, charts, and graphics. The Extract Images demo shows how to extract all embedded images from a PDF document using PDFium VCL, saving them in optimal formats based on their characteristics.

Overview

This demo extracts embedded images (bitmaps) from PDF pages and saves them as separate image files. It includes image preview, format detection, and batch extraction with progress tracking.

Key Features

  • Extract All Images – Extract every embedded image from the PDF
  • Page Range Selection – Extract from specific pages only
  • Smart Format Detection – Automatically choose JPEG, PNG, or BMP based on image characteristics
  • Image Preview – Preview extracted images before saving
  • Detailed Information – View dimensions, format, and file size
  • Batch Processing – Extract multiple images with progress tracking

PDFium DLL Requirements

Before running any PDFium VCL application, ensure the PDFium DLL files are installed:

  • pdfium32.dll / pdfium64.dll – Standard versions (~5-6 MB)
  • pdfium32v8.dll / pdfium64v8.dll – With V8 JavaScript engine (~23-27 MB)

Installation: Run PDFiumVCL\DLLs\CopyDlls.bat as Administrator to automatically copy the DLLs to Windows system directories.

Basic Image Extraction

Access embedded images through the Bitmap and BitmapCount properties:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
procedure ExtractImagesSimple;
var
  Pdf: TPdf;
  I, J: Integer;
  Bitmap: TBitmap;
begin
  Pdf := TPdf.Create(nil);
  try
    Pdf.FileName := 'document.pdf';
    Pdf.Active := True;
    
    // Loop through all pages
    for I := 1 to Pdf.PageCount do
    begin
      Pdf.PageNumber := I;
      
      // Loop through all images on this page
      for J := 0 to Pdf.BitmapCount - 1 do
      begin
        Bitmap := Pdf.Bitmap[J];
        try
          // Save as BMP
          Bitmap.SaveToFile(Format('Page%d_Image%d.bmp', [I, J + 1]));
        finally
          Bitmap.Free;
        end;
      end;
    end;
    
  finally
    Pdf.Active := False;
    Pdf.Free;
  end;
end;

Complete Extraction with Format Detection

The demo implements intelligent format selection:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
procedure TFormMain.ButtonExtractClick(Sender: TObject);
var
  I, J, StartPage, EndPage: Integer;
  Bitmap: TBitmap;
  FullFileName, DetectedFormat, ImageExtension: string;
  ImageInfo: TImageInfo;
begin
  FCancelled := False;
  FProcessedImages := 0;
  FTotalImages := 0;
  
  ClearExtractedImages;
  EnableControls(False);
  
  try
    Pdf.FileName := EditPdfFile.Text;
    Pdf.PageNumber := 0;
    Pdf.Active := True;
    
    ParsePageRange(EditPageRange.Text, StartPage, EndPage);
    if EndPage = -1 then
      EndPage := Pdf.PageCount;
      
    // Calculate total images for progress
    for I := StartPage to EndPage do
    begin
      Pdf.PageNumber := I;
      FTotalImages := FTotalImages + Pdf.BitmapCount;
    end;
    
    ProgressBar.Max := FTotalImages;
    
    // Extract images
    for I := StartPage to EndPage do
    begin
      if FCancelled then
        Break;
        
      Pdf.PageNumber := I;
      
      for J := 0 to Pdf.BitmapCount - 1 do
      begin
        if FCancelled then
          Break;
          
        Bitmap := Pdf.Bitmap[J];
        if Assigned(Bitmap) then
        begin
          try
            // Detect optimal format
            DetectedFormat := DetectImageFormat(Bitmap);
            ImageExtension := GetExtensionForFormat(DetectedFormat);
            
            FullFileName := Format('%s\Page%d_Image%d%s',
              [FCurrentOutputDir, I, J + 1, ImageExtension]);
              
            SaveBitmapInOptimalFormat(Bitmap, FullFileName);
            
            // Store image info for preview
            ImageInfo.FileName := FullFileName;
            ImageInfo.PageNumber := I;
            ImageInfo.ImageIndex := J + 1;
            ImageInfo.Width := Bitmap.Width;
            ImageInfo.Height := Bitmap.Height;
            ImageInfo.Format := DetectedFormat;
            ImageInfo.Bitmap := TBitmap.Create;
            ImageInfo.Bitmap.Assign(Bitmap);
            
            AddImageInfo(ImageInfo);
            
            Inc(FProcessedImages);
            ProgressBar.Position := FProcessedImages;
            
          finally
            Bitmap.Free;
          end;
        end;
      end;
    end;
    
    UpdateImageList;
    
  finally
    Pdf.Active := False;
    EnableControls(True);
  end;
end;

Intelligent Format Detection

Choose the optimal format based on image characteristics:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
function TFormMain.DetectImageFormat(ABitmap: TBitmap): string;
begin
  // Check if image has transparency (alpha channel)
  if ABitmap.PixelFormat = pf32bit then
  begin
    // PNG for transparency support
    Result := 'PNG';
  end
  // Check if it's likely a photographic image
  else if (ABitmap.Width * ABitmap.Height > 100000) and
          (ABitmap.PixelFormat in [pf24bit, pf32bit]) then
  begin
    // Large, complex image - use JPEG for smaller file size
    Result := 'JPEG';
  end
  else
  begin
    // Small or simple image - preserve quality with BMP
    Result := 'BMP';
  end;
end;
 
function TFormMain.GetExtensionForFormat(const AFormat: string): string;
begin
  case UpperCase(AFormat)[1] of
    'J': Result := '.jpg';
    'P': Result := '.png';
    'B': Result := '.bmp';
  else
    Result := '.bmp';
  end;
end;

Saving in Optimal Format

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
procedure TFormMain.SaveBitmapInOptimalFormat(ABitmap: TBitmap;
  const AFileName: string);
var
  JpegImg: TJPEGImage;
  FileExt: string;
begin
  FileExt := UpperCase(ExtractFileExt(AFileName));
  
  if FileExt = '.JPG' then
  begin
    // Save as JPEG with good quality
    JpegImg := TJPEGImage.Create;
    try
      JpegImg.Assign(ABitmap);
      JpegImg.CompressionQuality := 85; // Good quality/size balance
      JpegImg.SaveToFile(AFileName);
    finally
      JpegImg.Free;
    end;
  end
  else if FileExt = '.PNG' then
  begin
    // PNG would require additional library
    // Fall back to BMP for compatibility
    ABitmap.SaveToFile(ChangeFileExt(AFileName, '.bmp'));
  end
  else
  begin
    // BMP - lossless quality
    ABitmap.SaveToFile(AFileName);
  end;
end;

Using TPdfImage for Raw Image Data

For advanced use cases, access raw image data:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
procedure ProcessRawImageData;
var
  Pdf: TPdf;
  I: Integer;
  PdfImage: TPdfImage;
begin
  Pdf := TPdf.Create(nil);
  try
    Pdf.FileName := 'document.pdf';
    Pdf.Active := True;
    Pdf.PageNumber := 1;
    
    for I := 0 to Pdf.ImageCount - 1 do
    begin
      PdfImage := Pdf.Image[I];
      
      // Access raw image properties
      ShowMessage(Format('Image %d: %d x %d, %d bytes',
        [I, PdfImage.Width, PdfImage.Height, Length(PdfImage.Data)]));
        
      // PdfImage.Data contains raw pixel data
    end;
    
  finally
    Pdf.Active := False;
    Pdf.Free;
  end;
end;

Displaying Image Information

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
procedure TFormMain.UpdatePreview(Index: Integer);
var
  Info: TImageInfo;
begin
  if (Index >= 0) and (Index < Length(FExtractedImages)) then
  begin
    Info := FExtractedImages[Index];
    
    // Update preview
    if Assigned(Info.Bitmap) then
      ImagePreview.Picture.Assign(Info.Bitmap);
      
    // Update info display
    MemoInfo.Lines.Clear;
    MemoInfo.Lines.Add('File: ' + ExtractFileName(Info.FileName));
    MemoInfo.Lines.Add('Page: ' + IntToStr(Info.PageNumber));
    MemoInfo.Lines.Add('Dimensions: ' + IntToStr(Info.Width) +
                       ' x ' + IntToStr(Info.Height));
    MemoInfo.Lines.Add('Format: ' + Info.Format);
    if Info.Size > 0 then
      MemoInfo.Lines.Add('Size: ' + FormatFloat('#,##0', Info.Size) + ' bytes');
  end;
end;

Page Range Parsing

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
procedure TFormMain.ParsePageRange(const ARange: string;
  var AStartPage, AEndPage: Integer);
var
  RangeStr: string;
  DashPos: Integer;
begin
  RangeStr := Trim(ARange);
  AStartPage := 1;
  AEndPage := -1; // -1 means extract to end
  
  if (RangeStr = '') or (UpperCase(RangeStr) = 'ALL') then
    Exit;
    
  DashPos := Pos('-', RangeStr);
  if DashPos > 0 then
  begin
    // Range format: start-end
    AStartPage := StrToIntDef(Trim(Copy(RangeStr, 1, DashPos - 1)), 1);
    AEndPage := StrToIntDef(Trim(Copy(RangeStr, DashPos + 1, Length(RangeStr))), -1);
  end
  else
  begin
    // Single page
    AStartPage := StrToIntDef(RangeStr, 1);
    AEndPage := AStartPage;
  end;
end;

Open Extracted Images Folder

1
2
3
4
5
6
7
procedure TFormMain.ButtonOpenFolderClick(Sender: TObject);
begin
  if DirectoryExists(FCurrentOutputDir) then
    ShellExecute(Handle, 'open', PChar(FCurrentOutputDir), nil, nil, SW_SHOWNORMAL)
  else
    ShowMessage('Output directory does not exist.');
end;

Use Cases

  • Digital Asset Extraction – Extract photos and graphics from marketing materials
  • Document Conversion – Prepare images for web or other formats
  • Archive Processing – Extract images from scanned document archives
  • Content Analysis – Extract images for machine learning or analysis

Conclusion

The Extract Images demo shows how easy it is to extract embedded images from PDF documents with PDFium VCL. The component handles the complex PDF parsing while you focus on how to use the extracted images in your application.

Combined with intelligent format detection, you can build professional image extraction tools that produce optimised output for any use case.

Explore PDFium Component at loslab.com and unlock the content in your PDF documents.