Design and Implementation of a Lightweight Invoice OCR and Data Extraction System Using Tesseract

Invoices are commonly received by businesses as scanned documents, photographs or image files. Entering details manually into accounting systems is time-consuming and can lead to errors. This paper presents the design and preliminary implementation of a lightweight invoice OCR and data extraction system that aims to reduce manual effort. The proposed system uses OpenCV for image preprocessing, Tesseract OCR for text recognition, and regular expressions with keyword-based rules to extract important invoice details, including the invoice number, date, vendor name, and total amount. The extracted information is checked using basic validation rules and can be stored in CSV or SQLite format. A web interface is also being developed to allow users to upload invoices, review extracted details, correct errors, and export the results. Three configurations are proposed to compare OCR performance on original images, images with basic preprocessing, and images with additional skew correction. At the current stage, the upload interface and file validation have been implemented, while the main extraction modules and comparative testing remain under development. Therefore, this study presents the system design and planned evaluation rather than claiming measured accuracy results.