Extraction of reference lines and items from form document images with complicated background

摘要

The extraction of reference lines and items is a fundamental and crucial task in form document analysis. Most of the studies performed so far were done in connection with binary images. This paper proposes a method of extracting lines from gray-level images, by constructing a 2D pseudo Gaussian–Coiflet wavelet with adjustable rectangular support. We also present a method of extracting items using the extracted reference lines and multiresolution wavelet sub-images, which is independent of the intensity of the strokes and backgrounds. The experimental results demonstrate the effectiveness of our proposed methods.