Multi-oriented Handwritten Annotations Extraction from Scanned Documents

Abstract

In this paper, we present an integrated system able to localize multi-oriented handwritten annotations in scanned documents. Unlike previous single methods which limit colors or types of annotations to be extracted, the proposed method attempts to extract annotations by fusing three feature extraction techniques based on internal and external shape analysis. Our method consists of two processes: 1) a coarse segmentation process which divides the scanned document into text and non-text regions. 2) A fine segmentation process which consists of three steps: a feature extraction process, a classification process and a majority voting process which identifies the segmented regions as machine-printed or handwritten annotations. We find that our adaptive method outperform all individual methods. Experimental results on a set of 301 annotated scanned documents are reported.