Tesseract takes much time when processing image

[paulius-petkus]

Current Behavior

Tesseract stucks (hangs) when processing specific file with with page segmentation mode: "SparseText".

Command:

"C:\Program Files\Tesseract-OCR\tesseract.exe" page.png page --psm 11 -l eng

Test File:
https://drive.google.com/file/d/1zg0nHfiudF89PbER12U1Gi3lQzBcyoiw/view?usp=sharing

For reference reasons I am attaching original PDF (cause my goal is to do PDF OCR):
53402_01040415_doc002.pdf

Expected Behavior

Processing should perform fast like with segmentation mode: "Auto".

Suggested Fix

No response

tesseract -v

5.5.0.20241111

Operating System

Windows 11

Other Operating System

No response

uname -a

No response

Compiler

tesseract --version:

tesseract v5.5.0.20241111
leptonica-1.85.0
libgif 5.2.2 : libjpeg 8d (libjpeg-turbo 3.0.4) : libpng 1.6.44 : libtiff 4.7.0 : zlib 1.3.1 : libwebp 1.4.0 : libopenjp2 2.5.2
Found AVX2
Found AVX
Found FMA
Found SSE4.1
Found libarchive 3.7.7 zlib/1.3.1 liblzma/5.6.3 bz2lib/1.0.8 liblz4/1.10.0 libzstd/1.5.6
Found libcurl/8.11.0 Schannel zlib/1.3.1 brotli/1.1.0 zstd/1.5.6 libidn2/2.3.7 libpsl/0.21.5 libssh2/1.11.0

CPU

intel i7-13700H

Virtualization / Containers

No response

Other Information

No response

[stweil]

tesseract page.png page does not find any text, but tesseract page.png page -c thresholding_method=1 does after about 12 seconds.

I assume that page segmentation mode 11 simply takes a lot more time. With thresholding method 1 it terminates after 3:33 minutes in my test.

[paulius-petkus]

Hi @stweil ,

Thanks for the tip - I’m relatively new to the CLI, so I wanted to verify I’m doing exactly what you described.

Does setting thresholding_method=1 in your wrapper correspond to using Leptonica’s Otsu binarization step (the default in my installation) before OCR? In other words, is that the same as running:

"C:\Program Files\Tesseract-OCR\tesseract.exe" page.png page --psm 11 -l eng

with no additional thresholding flags?

When I run that exact command with this specific PNG, Tesseract hangs for about 20 minutes before I kill it. If I downsample the image by 4×, it completes in seconds. I also tried --psm 3 (Auto) on the full image, and it runs fast - but I’m reluctant to switch modes globally since it could hang on some other files where “SparseText” works fine. (don't want to change mode blindly)

Let me know if there are any extra options you’d recommend to make the raw --psm 11 run reliably and avoid such hanging.

[stweil]

The default is --thresholding_method=0 (Tesseract's own implementation of Otsu). And your original command finishs after 23:27 minutes on my MacBook, so Tesseract does not hang, but simply takes much time. I have images which take more than an hour (especially with very large tables).

[amitdo]

https://tesseract-ocr.github.io/tessdoc/ImproveQuality.html#binarisation

Tesseract 5.0.0 added two new Leptonica based binarization methods: Adaptive Otsu and Sauvola. Use tesseract --print-parameters | grep thresholding_ to see the relevant configurable parameters.