I'm trying to use Rmagick and Tesseract for OCR. I'm creating a simple image file with some text here:
canvas = Magick::ImageList.new
canvas.new_image(300, 300) { self.density = "500" }
text = Magick::Draw.new
text.annotate(canvas, 0,0,2,2, 'some_text') {
self.font = font_path
self.gravity = Magick::CenterGravity
self.pointsize = 100
self.density = '300'
}
canvas.write('tmp_text_img.png')
And I try to read it with a shell script here:
`tesseract #{input} tmp_text_from_img`
However, Tesseract keeps giving me a warning:
Warning. Invalid resolution 0 dpi. Using 70 instead.
This results in really crappy accuracy. Which I find strange because I'm explicitly setting it twice when I create the image. Is there something I'm doing wrong? Or is there a way to force tesseract to use the proper dpi?
Thanks all!
Solved it with:
canvas.write('tmp_text_img.png') {self.units= Magick::PixelsPerInchResolution; self.density = "300"}