Does not extract images from PDF

[robertovaldesperez]

Hello @jafin @ststeiger @saikatguha @nils-a @HakanL , I am trying to extract the images from this pdf but nothing is extracted, can you help me?

File: GLOVAL 03.pdf

[Bogdancev]

Hello, @robertovaldesperez

I'm not sure if it helps, but this is how I do it. I have a pdf-file which is produced by a scanner, that file contains several pages and each page has an image. This image can be jpeg or tif (I think your format "FlatDecode" is png, if I'm not mistaken. And if so, the library probably doesn't have a straight way of converting it).

1. File is already in memory stream, I go to specified page and extract the image:

   public static MemoryStream ExtractPageImage(MemoryStream ms, int PageNumber)
   {
    if (null == ms || 0 == ms.Length) return null;
    ms.Position = 0;
   
    PdfDocument document = PdfReader.Open(ms, PdfDocumentOpenMode.ReadOnly);
    if (document.Pages.Count <= PageNumber) return null;
    PdfPage pdfPage = document.Pages[PageNumber];
   
    PdfDictionary resources = pdfPage.Elements.GetDictionary("/Resources");
    if (resources != null)
    {
        PdfDictionary xObjects = resources.Elements.GetDictionary("/XObject");
        if (xObjects != null)
        {
            ICollection<PdfItem> items = xObjects.Elements.Values;
            foreach (PdfItem item in items)
            {
                if (item is PdfReference reference)
                {
                    if (reference.Value is PdfDictionary xObject && xObject.Elements.GetString("/Subtype") == "/Image")
                    {
                        var msImage = ExportImage(xObject);
                        return msImage;
                    }
                }
            }
        }
    }
    throw new Exception("image not found inside pdf-file.");
   }

2. Here is how I decide on the image format:

   static MemoryStream ExportImage(PdfDictionary image)
   {
    string filter = image.Elements.GetName("/Filter");
    MemoryStream msImage = filter switch
    {
        "/DCTDecode" => ExportJpegImage(image),
        "/CCITTFaxDecode" => ExportTifImage(image),
        _ => throw new Exception("\"" + filter + " is unsupported image format."),
    };
   
    Image img = Image.FromStream(msImage);
    if (img.Height > img.Width)
    {
        img.RotateFlip(RotateFlipType.Rotate90FlipNone); // Rotates the image clockwise 90 degrees
        var msRotated = new MemoryStream();
        img.Save(msRotated, System.Drawing.Imaging.ImageFormat.Jpeg);
        msImage.Dispose();
        return msRotated;
    }
    return msImage;
   }

3. And here is a methods for jpeg and tiff:

   
   static MemoryStream ExportJpegImage(PdfDictionary image)
   {
   try
   {
       // Fortunately JPEG has native support in PDF and exporting an image is just writing the stream to a file.
       byte[] stream = image.Stream.Value;
       var ms = new MemoryStream(stream);
       return ms;
   }
   catch
   {
       throw new Exception("failed converting JPEG image inside PDF-file.");
   }
   }

static MemoryStream ExportTifImage(PdfDictionary image) { try { MemoryStream ms = new MemoryStream();

    int width = image.Elements.GetInteger(PdfImage.Keys.Width);
    int height = image.Elements.GetInteger(PdfImage.Keys.Height);
    int bpp = image.Elements.GetInteger(PdfImage.Keys.BitsPerComponent);
    int K = 0;

    //https://github.com/gheeres/PDFSharp.Extensions/blob/master/Pdf/PdfDictionaryExtensions.cs - some explanations about DecodeParams and more
    PdfObject oDecodeParams = image.Elements.GetObject(PdfImage.Keys.DecodeParms);
    if (oDecodeParams is PdfDictionary)
        if (((PdfDictionary)oDecodeParams).Elements.ContainsKey("/K"))
            K = ((PdfDictionary)oDecodeParams).Elements.GetInteger("/K");

    byte[] data = image.Stream.Value;

    Tiff tiff = Tiff.ClientOpen("in-memory", "w", ms, new TiffStream());
    tiff.SetField(TiffTag.IMAGEWIDTH, (uint)(width));
    tiff.SetField(TiffTag.IMAGELENGTH, (uint)(height));
    if (0 > K)
        tiff.SetField(TiffTag.COMPRESSION, Compression.CCITTFAX4);
    if (0 <= K)
        tiff.SetField(TiffTag.COMPRESSION, Compression.CCITTFAX3);
    tiff.SetField(TiffTag.BITSPERSAMPLE, (uint)(bpp));
    tiff.SetField(TiffTag.SAMPLESPERPIXEL, 1);

    tiff.WriteRawStrip(0, data, data.Length);
    tiff.Flush();
    return ms;
}
catch
{
    throw new Exception("failed converting TIFF image inside PDF-file.");
}

}

I hope this helps, I did it 5 or 6 years ago and it still works perfectly.
Regards, Sergii

[robertovaldesperez]

Hi @Bogdancev, I'm going to try this code, Do you have an example of how to extract a PNG image?

[Bogdancev]

Hi @robertovaldesperez Unfortunally not, I have a full control over the settings of the company scanners, so they are set up to produce jpegs or tiffs.

[robertovaldesperez]

Hi, I had to create a recursive method to extract images, because some were inside other resources...

private static IEnumerable<PdfDictionary> GetXObjectImages(PdfDictionary pdfDictionary) { var resources = pdfDictionary.Elements.GetDictionary("/Resources"); if (resources != null) { var xObjects = resources.Elements.GetDictionary("/XObject"); if (xObjects != null) { foreach (var item in xObjects.Elements.Values) { if (item is PdfReference reference) { if (reference.Value is PdfDictionary xObject) { if (xObject.Elements.GetString(PdfImage.Keys.Subtype) == "/Image") { yield return xObject; } else { foreach (var xObject1 in GetXObjectImages(xObject)) { if (xObject1.Elements.GetString(PdfImage.Keys.Subtype) == "/Image") { yield return xObject1; } } } } } } } } }