Nezuyomi - OCR image viewer


This is basically a toy at this point, so don't expect it to work well or be easy to use.

Unlike other OCR programs, Nezuyomi can be used with any command line OCR system, and saves OCR results to disk.

- Image viewer with sharp (sinc/jinc) downscaling, edge enhancement, three zoom modes
- Uses modern OpenGL
- Pages through the directory of the image you opened it with, fast pages changes, appropriate for reading manga/comics as loose images
- Lets you define regions of text on the image and OCR them with an external OCR program, details in readme, or use "ocr easy"
- Three OCR settings, can do whatever you want (horizontal, vertical, english, autodetect, etc), see readme ("easy mode" settings are vertical japanese, horizontal japanese, and english)
- Regions can have defined skew and text scale to improve OCR accuracy, sent to the OCR script (see readme)
- Region text copied to clipboard and displayed as a subtitle when you click on the region
- Region text can be pasted back into the program after being edited
- Saves regions and region text to disk (under userdir/ネズヨミ/ on windows and userdir/.config/ネズヨミ on unix, needs to exist first or saving won't work) in a plain text format, can review them later or share them with other people

Not yet implemented:
- Only willing to open images with png or jpg extensions right now, sorry
- No CBZ/etc support, sorry
- No fine zoom yet, only scaling modes, sorry
- Only runs on machines that support OpenGL 3.3
- No real GUI (aside from the image display), the only controls are keyboard bindings. This isn't a design decision, it's me being lazy.

[Image: Xlevkgh.png]
Edited: 2017-10-09, 3:31 pm
Uploaded an update. This makes it easier get OCR working on windows. You no longer have to create a folder that starts with a dot, something windows explorer doesn't allow you to do.

To make OCR work, extract 0.1.10's into C:/Users/username/, so that the file C:/Users/username/ネズヨミ/ocr.txt exists. You also need to install imagemagick. If you're very good with computers you can set up your own OCR scripts instead.
Edited: 2017-10-06, 10:13 am