c++ - OpenCV Gaussian blur breaks Tesseract? -
the problem: week or ago, in hopes of improving tesseract's accuracy more, added guassian blur / otsu binarization combo results in beautiful binary images 1 attached. in opencv, image pass tesseract binary image (like 1 attached). when tesseract pre-processing of image (even 1 posted below) image becomes corrupted , therefore no meaningful output produced. see image below example input, idea of tesseract doing image.
the source of problem guassian blur. if remove it, thresholded image
tesseract outputs not garbled, not clean , readable binary image attached. can disable tesseract pre-processing images pass it? why guassian blur ruin tesseract? feel if input image clear 1 attached, accuracy improved.
both images of same column. first input image, second result of tesseract's image pre-processing.
input tesseract example:
tesseract corruption (obtained getthresholdedimage()
:
i suggest save image data tesseract (tess.getthresholdedimage()) , store disk after tess.setimage(), can sure provided correct image ocr.
Comments
Post a Comment