Extracting the attached PDF with the following command:
java -jar app/target/pdfbox-app-1.7.0-SNAPSHOT.jar PDFToImage -imageType png /tmp/WO2011140338-Page25.pdf
This generates image WO2011140338-Page251.png (attached).
This is a great improvement over pdfbox 1.6.0, which generates four copies of the image (maybe because of the CMYK encoding?). Well done!
However the image is still of quite poor quality, apparently lower than the actual image data in the PDF, when displayed with acrobat reader or evince 2.32.0 (screenshot attached too). It would be great if that could be fixed too.