Hi, On Wed, Apr 8, 2009 at 3:18 PM, fs <[email protected]> wrote:
> > Hey. > I'm currently adapting the text-line-recognizer to Japanese script. > Japanese documents often have mixed horizontal and vertical text > orientation in the same document. I have developed a extension of the > RAST-textline-recognizer to deal with the different orientations. Now > I'd like to evaluate the performance. I read the following paper: > 'Performance Evaluation and Benchmarking of Six Page Segmentation > Algorithms'. > I think I can use the same approach with some modifications. Is the > code for the comparison and ground-truthing tool somewhere available? The code for pixel-accurate evaluation is already in OCRopus (seg-eval.cc). The evaluation scheme of Mao is available as part of PSET toolkit: http://www.kanungo.com/software/pset-1.01.tar.gz For ground-truthing, we used Kolourpaint to develop line-level ground-truth followed by simple post-processing as described in the paper. > > I would also appreciate some hints on literature for layout analysis > of Japanese script, if you can think of any. Voronoi page layout analysis should work well on Japanse. An implementation is also part of OCRopus (SegmentPageByVoronoi). There is also an open source implementation of a Japanese layout anlysis in project O2: http://www.sc.isc.tohoku.ac.jp/~hgot/O2/ Cheers, Faisal > > > Best regards. > > > --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "ocropus" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/ocropus?hl=en -~----------~----~----~----~------~----~------~--~---
