Cleaning up an image for OCR with ImageMagick and 'textcleaner' - imagemagick

I have the following image that I'd like to prepare for an OCR with tesseract:
The objective is to clean up the image and remove all of the noise.
I'm using the textcleaner script that uses ImageMagick with the following parameters:
./textcleaner -g -e normalize -f 30 -o 12 -s 2 original.jpg output.jpg
The output is still not so clean:
I tried all kinds of variations for the parameters but with no luck.
Does anyone have an idea?

If you convert to JPEG, you will always have the type of artifacts you are seeing.
This is a typical "feature" of JPEG compression. JPEGs are never good for images showing sharp lines, contrasts with uniform colors between different areas of the image, using only very few colors. This is true for black + white texts. JPEG is only "good" for typical photos, with lots of different colors and shading...
Your problem will most likely completely get resolved if you use PNG as an output format. The following image demonstrates this. I generated it with the same parameters as your last example command used, but with PNG as the output format:
textcleaner -g -e normalize -f 30 -o 12 -s 2 \
http://i.stack.imgur.com/ficx7.jpg \
out.png
Here is a similar zoom into the output:
You can very likely improve the output even more if you play with the parameters of the textcleaner script. But that is your job... :-)

Related

How to create a gif from an image sequence without dither with FFMPEG?

I'm able to create a gif from the image sequence, but I'm struggling to remove the dither from it.
This is the command I'm using to create the gif:
ffmpeg -f image2 -framerate 24 -y -i image_%d.png -loop -1 "C:\Users\agetr\Documents\_tmp\__giuf.gif"
And I've tried to use the paletteuse=dither=none filter in different ways with no luck.
P.S.: I'm very new to the ffmpeg cli
You need to use -sws_dither none (after the -i $file argument, and before the output file). I've tried this on current git/master of FFmpeg and it works as expected, but on older builds (e.g. 4.4.1) this doesn't work. I don't know why exactly, so use a recent (5.0 or any version from 2022, if possible) version/build.

Batch-converting of INTA images

I have a large number of images in the INTA format, an old SGI standard. INTA is a grayscale image with an alpha channel. All of these need to be converted to TGA files. The problem is that neither ImageMagick nor PIL/Pillow seem to be able to parse them correctly. ImageMagick can read and export them but doesn't seem to understand the alpha channel, and PIL fails to open them, with the error ValueError: Unsupported SGI image mode. The one thing that I've found that reads them successfully is GIMP:
An ideal solution would be one that is easy to invoke from a script.
For reference, here is one of the images in question (the same one seen in the screenshot): https://www.dropbox.com/s/8hoppdgtuqxsy26/girder01.inta?dl=0
It seems GDAL is able to read your image and I converted it to a greyscale+alpha PNG using:
gdal_convert YOURIMAGE.sgi result.png
You can easily get to TGA from there.
I am assuming the batching is not an issue, but it would look something like this in bash:
mkdir -p OUTPUT
for f in *.inta ; do
gdal_translate "$f" OUTPUT/"$f"
done
I had all sorts of trouble installing GDAL on macOS so I just used docker like this:
docker run --rm -v /Users:/Users osgeo/gdal:alpine-normal-latest gdal_translate /Users/mark/Downloads/image.sgi /Users/mark/Downloads/result.png

How to batch convert JPG image gallery to BPG image format

How to convert whole image gallery or family album from JPG to BPG image format?
I'm looking for some batch conversion tool, application or script on Windows platform.
Input directory must be processed recursively and image quality should be preserved.
Linux command line fragment I use for this task, with current directory being gallery of '*.JPG' files, without subdirectories.
parallel -i sh -c 'convert -quality 100 {} -scale "1280x1000>" {}.png && bpgenc -q 30 {}.png -o {}.bpg && rm -f {}.png' -- *.JPG
You may adjust (or remove) resizing and change -q 30 to lower value for more quality.
It depends on ImageMagick and bpgenc.
To run in on Windows, you probably will need Cygwin.
Look at
http://www.romeolight.com/products/bpgconv/
for nice Windows converter.
2 things to mention: Currently there is options menu in top right of window. And all BPG pictures are saved into folder on your desktop called bpg_encoded.
Martin

How to flop multiple images using imagemagik?

I tried to flip horizontally the images in a specific folder using ImageMagik. But the mogrify -flop *jpg is changing all the images in their mirror images. I want to keep the initial images and for the flopped ones I want to rename them as *_flop.jpg. I am stuck: How to do it?
Assuming you are on Linux or OSX, like this:
#!/bin/bash
for f in *.jpg; do
new="${f%%.jpg}_flop.jpg"
echo convert "$f" -flop "$new"
done
At the moment, it does nothing, it just tells you what it would do. If you like what it shows, just remove the word echo and run it again.
Save the code above as flopper, and then go to Terminal and type this:
chmod +x flopper # Just do this one time to make the script executable (runnable)
./flopper # Actually run it
Update:
For newer versions of ImageMagick (version 7 onwards) you do not need to write:
convert -flop
Instead you need to write:
magick -flop
So your script essentially becomes:
#!/bin/bash/
for file in *.jpg
do
magick $file -flop ${file%%.jpg}_flop.jpg
done
Note: This works when your filenames don't have spaces or newlines in them.

How to convert an image to PVRTC format

I've been using terminal to convert my jpg image named left to pvrtc format and this happens:
/Developer/Platforms/iPhoneOS.platform/Developer/usr/bin/texturetool -e PVRTC --bits-per-pixel-2 -o left.pvrtc left.jpg
Failed to load image
Failed to perform Encode
Change:
/Developer/Platforms/iPhoneOS.platform/Developer/usr/bin/texturetool -e PVRTC --bits-per-pixel-2 -o left.pvrtc left.jpg
To:
/Developer/Platforms/iPhoneOS.platform/Developer/usr/bin/texturetool -e PVRTC --bits-per-pixel-2 -o left.pvr left.PNG
The file must be a PNG
Updated instructions on OSX 10.10+
In Terminal, hit cd to make sure you're in your root directory.
cd /Applications/Xcode.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/usr/bin/ && ls
This should show you a list of all the tools there (Make sure you see texturetool) - If you don't, update your OS, then Xcode
Once you know it's there, cd back to your root and run the tool like:
/Applications/Xcode.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/usr/bin/texturetool -e PVRTC --bits-per-pixel-2 -f PVR -o [new path to pvr to go] [path of original png]
Your output pvr path can be ~/whatever/image.pvr and your input png is the path to your image to be converted.
HERE and HERE are other good walkthrough
The iPhone SDK includes a tool that allows you to create textures in the PVRTC compression format, aptly named texturetool. If you have Xcode installed with the iPhone OS 2.2 SDK in the default location (/Developer/Platforms/), then texturetool is located at: /Developer/Platforms/iPhoneOS.platform/Developer/usr/bin/texturetool.
texturetool allows you to create four variants of PVRTC data, the primary difference being tradeoffs between quality and size. You will have to experiment with these variants to determine which setting is the best compromise for each individual texture image.
Apple Docs
HERE is some same code to help once you have your pvrtc file
HERE are some more docs
HERE is a good read on PVRTC stuff

Resources