Efficient Technique for Automatic Extraction and Identification of Text from an Image using Friend Pattern Chain and Euclidean Distance

A reference map acts as the fundamental source of information pertaining to various morphological features in studies related to Geographic information system (GIS). The feature set includes rivers, contours, transportation-network, national and international demarcation to name a few. Each of these features is represented on the reference map with the help of different color code to ease the task of visual interpretation while digitizing them. So reference map can be considered as superimposed layers of information casted on the terrain. One of the key features associated with a reference map is text demarking the landmarks and the rivers etc. These texts play an important role in identifying the land marks and rivers while performing studies related to drainage network or demography. These features are can possibly digitized in two ways either using traditional manual techniques which involves visually interpreting individual characters and then grouping it to form a name. This process takes demands increased effort, time and precision from the researchers responsible for digitizing the feature. Instead relying on the less effective traditional manual approach an efficient automatic extraction procedure can be developed that takes into consideration certain pre acquired knowledge for identifying the text. This proposed work introduces a rotation invariant text identifying procedure that relies on the combination of friend pattern chain concept and distance.