Identification of scripts and orientations of degraded document images

This paper presents a pair of identification technique that automatically detect scripts and orientations of document images suffering from various types of document degradation. In the proposed technique, scripts and orientations of document images are determined through the document vectorization, which converts a text image into a pair of document vectors that characterize stroke density and stroke distribution, respectively. For each script under study at each of the two typical orientations (upright and upside-down), a number of reference document vectors are first constructed. Script and orientation of the query document are then determined according to the distances between the query document vectors and the pre-constructed reference document vectors. Experiments show that the proposed techniques are accurate and tolerant to various types of document degradation. Key words: Document orientation identification, document script identification, document image analysis.

• Type: Journal
• Year: 2010
• Where: PAA
• Authors: Shijian Lu, Linlin Li, Chew Lim Tan
• Keywords: Distributed And Parallel Computing, Document, Document Degradation, Document Vectors, Operations Research, PAA 2010