Detection Masking for Improved OCR on Noisy Documents

TL;DR

This paper introduces a masking-enhanced detection network that filters non-textual elements to improve OCR accuracy on noisy, degraded documents, leveraging contextual information and a new synthetic dataset.

Contribution

It proposes a novel detection network with masking to enhance OCR on degraded documents and provides a synthetic dataset for improved detection training.

Findings

Enhanced OCR accuracy on noisy documents

Effective filtering of non-textual elements

Broad applicability demonstrated on public dataset

Abstract

Optical Character Recognition (OCR), the task of extracting textual information from scanned documents is a vital and broadly used technology for digitizing and indexing physical documents. Existing technologies perform well for clean documents, but when the document is visually degraded, or when there are non-textual elements, OCR quality can be greatly impacted, specifically due to erroneous detections. In this paper we present an improved detection network with a masking system to improve the quality of OCR performed on documents. By filtering non-textual elements from the image we can utilize document-level OCR to incorporate contextual information to improve OCR results. We perform a unified evaluation on a publicly available dataset demonstrating the usefulness and broad applicability of our method. Additionally, we present and make publicly available our synthetic dataset with a unique hard-negative component specifically tuned to improve detection results, and evaluate the benefits that can be gained from its usage