Tesseract: Are there more PSM modes than are listed in the help/wiki - 11 and 12?

[samiles]

Hi,

The command line help output shows 11 PSM modes (0 through 10).

  pagesegmode values are:
  0 = Orientation and script detection (OSD) only.
  1 = Automatic page segmentation with OSD.
  2 = Automatic page segmentation, but no OSD, or OCR
  3 = Fully automatic page segmentation, but no OSD. (Default)
  4 = Assume a single column of text of variable sizes.
  5 = Assume a single uniform block of vertically aligned text.
  6 = Assume a single uniform block of text.
  7 = Treat the image as a single text line.
  8 = Treat the image as a single word.
  9 = Treat the image as a single word in a circle.
  10 = Treat the image as a single character.

I was trying each one and getting mixed results. However, I accidentally ran 'psm -11' and I suddenly got perfect accuracy - way way better than any other PSM mode, and much better than the default. The same for PSM 12 too, perfect accuracy - then PSM 13 gives nothing.

The image is just about 10 words over 2 lines, spread about the page. All the other segmentation modes and default garble the text, but PSM 11/12 worked great, splitting text perfectly.

Is it correct that there's a PSM 11 and 12 mode? What do they do, why do they give such good accuracy?! And should they be in the help/Wiki?

Thanks!

[amitdo]

Is it correct that there's a PSM 11 and 12 mode?

Yes. and PSM 13 too.
You revealed our secret! :)
https://github.com/tesseract-ocr/tesseract/blob/8d6dbb133b41/api/tesseractmain.cpp#L115

I'm sorry, I do not have good answers to your other questions.

[amitdo]

With Tesseract 4.0 PSM 11, 12, and 13 appear in the help message. psm 13 is used with the new LSTM engine to OCR a single textline image.

[zdenop]

master(4.00) & 3.05 repository produce help message for psm 11-13

[crifan]

for

●✚  tesseract --version 
tesseract 4.1.0
 leptonica-1.78.0
  libgif 5.2.1 : libjpeg 9c : libpng 1.6.37 : libtiff 4.1.0 : zlib 1.2.11 : libwebp 1.0.3 : libopenjp2 2.3.1
 Found AVX2
 Found AVX
 Found SSE

support psm=11 -13:

 tesseract --help-psm
Page segmentation modes:
  0    Orientation and script detection (OSD) only.
  1    Automatic page segmentation with OSD.
  2    Automatic page segmentation, but no OSD, or OCR. (not implemented)
  3    Fully automatic page segmentation, but no OSD. (Default)
  4    Assume a single column of text of variable sizes.
  5    Assume a single uniform block of vertically aligned text.
  6    Assume a single uniform block of text.
  7    Treat the image as a single text line.
  8    Treat the image as a single word.
  9    Treat the image as a single word in a circle.
 10    Treat the image as a single character.
 11    Sparse text. Find as much text as possible in no particular order.
 12    Sparse text with OSD.
 13    Raw line. Treat the image as a single text line,
       bypassing hacks that are Tesseract-specific.