Document image processing: graphics recognition

Document analysis or processing is mainly related to texts and graphics. It concerns separation, localisation and recognition. According to Nagy (2000), document analysis is related to document image analysis (DIA) since the overall research works have been concerned with document image interpretation. In a similar manner, Kasturi et al. (2002) categorises document image analysis into two domains: 1) textual processing and 2) graphical processing. Basic tasks are image segmentation, layout understanding and graphical symbol recognition. More specifically, graphics processing refers to lines, corners and curves analysis on the one side while on the other side, it is concerned with regions and symbol analysis and recognition. One part of DIA means to recognise graphical elements and or to localise significant or known visual part of the images. Within the aforementioned context, graphics recognition (i.e., recognition of graphical symbols) has been an intensive research activity in the community of pattern recognition and document analysis since a few decades. Symbol recognition has a long history since the 70’s. In 1998, Tombre (1998) has made a statement: ‘none of these methods works’ in general. Since then, it has been actively extended.