Text extraction from scanned land map images

The information in a document comes in many forms; this includes texts, tables, charts, symbols, maps, logos, stamps, photographs etc. All these forms may be combined in suitable forms to create a document. Paper land maps are useful documents for collection of geographical information. A land map image can be considered as a complex document image as the texts contained there may have a complex background consists of various intensity values and varied orientations. This paper presents two methods for text extraction from scanned land map images. In the first method we have used mathematical morphological operations for making the symbols of text be connected. Then mean and standard deviations are used for text non-text separation. In the second method we have extracted texts based on intensity analysis. The approaches are tested on a collected dataset of map images and experimental results are encouraging.