European Script Identification using Projection Profiles and Statistical Features for Handwritten Documents
Aditi Gaygole, Dinesh Vitthalrao Rojatkar · 2019
Europe has rich history of Roman Empire and many more but the biggest problem to extract this information is the lack of knowledge of the languages that were used at that time. There are almost 17 scripts[1] that were used in European countries in different duration of time which covers hundreds of languages. In order to preserve these this old document it is very important to digitized these documents. Manual digitization of these document is very difficult as there are very few people who understand these languages and also manual digitization is time consuming process. Hence this paper explain texture based approach for 14 different European script identification. Here we have used 22 features on the basis of GLRLM, GLCM, projection profiles and geometric moment features. Classification is done using SVM classifier and we obtain overall 90.8% accuracy.