Identify text area in image and crop it in R

[rigel_1997]

Is there a way in R (R Studio) to identify the text area in the image and crop it? Like I need to locate the license plate only and crop it then extract the text using tesseract. Thank you.

[technocrat]

It's going to be difficult, as shown in the sample below, trying OCR on the full image, an image cropped programmatically but with height/width and cropped area set by hand and of a manual screenshot.

library(tesseract)
eng <- tesseract("eng")
text <- tesseract::ocr("https://forum.posit.co/uploads/default/original/3X/d/5/d591d3342bcc274fcb16d78924d65f9547a25d50.png", engine = eng)
cat(text)
#> oF
#> — —

library(magick)
#> Linking to ImageMagick 6.9.11.60
#> Enabled features: fontconfig, freetype, fftw, heic, lcms, pango, webp, x11
#> Disabled features: cairo, ghostscript, raw, rsvg
#> Using 12 threads
img <- image_read("https://forum.posit.co/uploads/default/original/3X/d/5/d591d3342bcc274fcb16d78924d65f9547a25d50.png")
img

cropped <- image_crop(img, "320x100+150+100")
cropped

tesseract::ocr(cropped, engine = eng)
#> [1] ""

shot <- image_read("~/Desktop/Screenshot from 2022-01-01 11-09-34.png")
shot

tesseract::ocr(shot, engine = eng)
#> [1] "WOR SIGK |\n"