1

I am have a bit of a problem converting some color PDFs to tiff images. The PDFs I am having problems with have hand written signatures written in blue ink. These signatures do not appear in the generated binary tiffs. I suspect there is a threshold value somewhere to determine which pixels will be black and which will be white.

@SuppressWarnings("serial")
private static void convertPdfToTiff(final File pdf, final File tif) throws Exception {
    try 
    {
        final Iterator<ImageWriter> imageWriterIterator = ImageIO.getImageWritersByFormatName("TIF");
        final ImageWriter imageWriter = imageWriterIterator.hasNext() ? imageWriterIterator.next() : null;

        final TIFFImageWriteParam writeParam = new TIFFImageWriteParam(Locale.getDefault());
        writeParam.setCompressionMode(TIFFImageWriteParam.MODE_EXPLICIT);
        writeParam.setCompressionType("LZW");

        PDDocument pdfDocument = PDDocument.load(pdf);
        PDFRenderer pdfRenderer = new PDFRenderer(pdfDocument);

        OutputStream out = new FileOutputStream(tif);
        final BufferedOutputStream bufferedOutputStream = new BufferedOutputStream(out);
        final ImageOutputStream imageOutputStream = ImageIO.createImageOutputStream(bufferedOutputStream);
        imageWriter.setOutput(imageOutputStream);
        imageWriter.prepareWriteSequence(null);

        int pageCounter = 0;
        for (PDPage page : pdfDocument.getPages()) 
        {
            BufferedImage image = pdfRenderer.renderImageWithDPI(pageCounter, 300, ImageType.BINARY);

            final IIOImage s = new IIOImage(image, null, new TIFFImageMetadata(new TIFFIFD(new Vector<BaselineTIFFTagSet>() 
            {
                {
                       add(BaselineTIFFTagSet.getInstance());
                }
            }))) 

            {   
                {
                       final TIFFImageMetadata tiffMetadata = (TIFFImageMetadata) getMetadata();
                       final TIFFIFD rootIFD = tiffMetadata.getRootIFD();
                       final BaselineTIFFTagSet base = BaselineTIFFTagSet.getInstance();
                       rootIFD.addTIFFField(new TIFFField(base.getTag(BaselineTIFFTagSet.TAG_X_RESOLUTION), TIFFTag.TIFF_RATIONAL, 1, new long[][] { { 300, 1 } }));
                       rootIFD.addTIFFField(new TIFFField(base.getTag(BaselineTIFFTagSet.TAG_Y_RESOLUTION), TIFFTag.TIFF_RATIONAL, 1, new long[][] { { 300, 1 } }));
                }
            };

            imageWriter.writeToSequence(s, writeParam);
            pageCounter++;
        }

        imageWriter.dispose();
        imageOutputStream.flush();
        imageOutputStream.close();
        bufferedOutputStream.flush();
        bufferedOutputStream.close();
        pdfDocument.close();
        out.flush();
        out.close();
    } 
    catch (Exception e) 
    {
        e.printStackTrace();
        throw e;
    }
}
Jeremy W
  • 1,751
  • 6
  • 25
  • 33
Safford96
  • 23
  • 7

2 Answers2

4

I had the same problem (blue signatures) some time ago and I did this:

  • render to RGB
  • convert to b/w with a filter from JH Labs (I got pointed to this by a comment in this answer)
  • I initially tried the dither and the diffusion filter
  • the filter that worked best for me was the bias part (I think I used 0.3) of the gain filter combined with the diffusion filter.
  • you can combine two filters with the compound filter.
  • the jhlabs stuff is not available as .jar file, but you can download the sources and add it to your project
  • some examples

Btw, save your files not as LZW, but as G4, that'll make them smaller. PDFBox has methods to efficiently save into images, see here. ImageIOUtil.writeImage() will save to G4 compressed TIFF if your BufferedImage is of type BITONAL.

Tilman Hausherr
  • 14,950
  • 6
  • 51
  • 80
  • Thanks for the response. That gave me the idea of rendering the image as ImageType.GRAY while create a new binary bufferedimage and drawing the gray image into the new binary image. – Safford96 May 04 '16 at 23:16
0

I ended up rendering the image as grayscale and re drawing it to a second bw image.

@SuppressWarnings("serial")
private static void convertPdfToTiff(final File pdf, final File tif) throws Exception {
    try 
    {
        final Iterator<ImageWriter> imageWriterIterator = ImageIO.getImageWritersByFormatName("TIF");
        final ImageWriter imageWriter = imageWriterIterator.hasNext() ? imageWriterIterator.next() : null;

        final TIFFImageWriteParam writeParam = new TIFFImageWriteParam(Locale.getDefault());
        writeParam.setCompressionMode(TIFFImageWriteParam.MODE_EXPLICIT);
        writeParam.setCompressionType("CCITT T.6");

        PDDocument pdfDocument = PDDocument.load(pdf);
        PDFRenderer pdfRenderer = new PDFRenderer(pdfDocument);

        OutputStream out = new FileOutputStream(tif);
        final BufferedOutputStream bufferedOutputStream = new BufferedOutputStream(out);
        final ImageOutputStream imageOutputStream = ImageIO.createImageOutputStream(bufferedOutputStream);
        imageWriter.setOutput(imageOutputStream);
        imageWriter.prepareWriteSequence(null);

        int pageCounter = 0;
        for (PDPage page : pdfDocument.getPages()) 
        {
            BufferedImage image = pdfRenderer.renderImageWithDPI(pageCounter, 300, ImageType.GRAY);
            BufferedImage image2 = new BufferedImage(image.getWidth(), image.getHeight(), BufferedImage.TYPE_BYTE_BINARY);
            Graphics2D g = image2.createGraphics();
            g.drawRenderedImage(image, null);
            g.dispose();

            final IIOImage s = new IIOImage(image2, null, new TIFFImageMetadata(new TIFFIFD(new Vector<BaselineTIFFTagSet>() 
            {
                {
                       add(BaselineTIFFTagSet.getInstance());
                }
            }))) 

            {   
                {
                       final TIFFImageMetadata tiffMetadata = (TIFFImageMetadata) getMetadata();
                       final TIFFIFD rootIFD = tiffMetadata.getRootIFD();
                       final BaselineTIFFTagSet base = BaselineTIFFTagSet.getInstance();
                       rootIFD.addTIFFField(new TIFFField(base.getTag(BaselineTIFFTagSet.TAG_X_RESOLUTION), TIFFTag.TIFF_RATIONAL, 1, new long[][] { { 300, 1 } }));
                       rootIFD.addTIFFField(new TIFFField(base.getTag(BaselineTIFFTagSet.TAG_Y_RESOLUTION), TIFFTag.TIFF_RATIONAL, 1, new long[][] { { 300, 1 } }));
                }
            };

            imageWriter.writeToSequence(s, writeParam);
            pageCounter++;
        }

        imageWriter.dispose();
        imageOutputStream.flush();
        imageOutputStream.close();
        bufferedOutputStream.flush();
        bufferedOutputStream.close();
        pdfDocument.close();
        out.flush();
        out.close();
    } 
    catch (Exception e) 
    {
        e.printStackTrace();
        throw e;
    }
}
Safford96
  • 23
  • 7