An Automation of invoice examination under letter of credit based on UCP 600 using OCR and rule-based text matching

Avyodri, Ridvy (2022) An Automation of invoice examination under letter of credit based on UCP 600 using OCR and rule-based text matching. Masters thesis, Universitas Pelita Harapan.

Preview	Text (Title) Title.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (69kB) | Preview
Preview	Text (Abstract) Abstract (2)_watermark.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (260kB) | Preview
Preview	Text (ToC) ToC_watermark.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (823kB) | Preview
Preview	Text (Chapter 1) Chapter1_watermark.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (1MB) | Preview
	Text (Chapter 2) Chapter2_watermark.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (2MB)
	Text (Chapter 3) Chapter 3_watermark.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (2MB)
	Text (Chapter 4) Chapter 4_watermark.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (450kB)
	Text (Chapter 5) Chapter 5_watermark.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (322kB)
Preview	Text (Bibliography) Bibliography_watermark.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (602kB) | Preview
	Text (Appendices) Binder4.pdf Restricted to Repository staff only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (4MB)

Abstract

Letter of Credit (L/C) is a trade finance product that is popularly used worldwide. L/C is a conditional guarantee of payment from issuing bank to the seller. The condition that has to be met is presentation of documents which complies with the L/C and every applicable rule such as UCP 600. Invoice is one of the most common documents that mostly required in L/C conditions. In current practices, compliance of the invoices is examined manually. These practices affect the accuracy, cost, and time efficiency of the examination considering the high L/C transaction volume. The needs to automate this examination process appear due to this background. This research proposed a tools pipeline that automate the invoice examination process. Three main processes proposed are; Text data extraction from L/C, Data extraction from invoice, and rule-based matching between pre-extracted data. Several techniques are used for the proposed processes namely, rule-based named entity for text data extraction from L/C, connected component analysis for determination of signature location, Hough transform for de-skewing the image, and Gaussian blur for minimization noise. Tesseract OCR 5.0 is used for OCR engines. To test the effectiveness of proposed method, experiment is done in 3 processes. First, text data extraction from L/C. Dataset used in this step are 47 L/Cs in text format. Second, data extraction from invoice images. Dataset used in this step are 16 invoices with 13 different formats. Third, rule-based matching between previously extracted data. Dataset used in this step are extracted data from previous step. Experiments results are 100% accuracy for text data extraction from L/C using gazetteers, average of 90.62% accuracy for data extraction from invoice images and average of 83.75% accuracy for rule-based matching between preextracted data.

Actions (login required)

View Item