Skip to content
This repository has been archived by the owner on Feb 11, 2024. It is now read-only.

Latest commit

 

History

History
61 lines (36 loc) · 1.89 KB

nshuti2015mobile.md

File metadata and controls

61 lines (36 loc) · 1.89 KB
Raw

Mobile Scanner and OCR (A first step towards receipt to spreadsheet)

Clement Ntwari Nshuti

Browse

@inproceedings{nshuti2015mobile,
  title={Mobile Scanner and OCR ( A first step towards receipt to spreadsheet )},
  author={Nshuti, Clement Ntwari},
  year={2015}
}

Pipeline

Receipt detection Receipt localization Receipt normalization Text line segmentation Optical character recognition Semantic analysis
❌ ✔️ ✔️ ❌ ❗ ❌

Receipt localization

  • denoised using a gaussian blur

  • sharpened

  • a canny edge detector with [75, 200] as the thresholds

  • extract the largest quadrilateral by approximating all closed regions in the output of the edge detector by quadrilaterals and keeping the largest from these.

Receipt normalization

  • This quadrilateral is the extracted from the original (undenoised and unsharpened) image and warped into a straight rectangle.

  • binarized using adaptive threshold

Optical character recognition

  • Tesseract

Notes

  • 7 conditions tested:

    • DARK.

      The best conditions underwhich a picture can be taken is from the top with a dark background. This will be used as the baseline. Other configurations will be just a variation from this. We’ll either vary the background color, the camera orientation or the document quality.

    • BRIGHT.

      Pictures of the document from the top with a bright background

    • NOISY

      Pictures of the document with a background that has several small patterns.

    • SIDE

    • FRONT

    • FOLDED

    • CRINKLED