Place: Large Lecture Room – CVC
Affiliation: Information and Communication Systems Engineering department. University of the Aegean. Greece
The presentation will include common tasks of document image processing as binarization, show-through noise removal, ruling line removal and text line segmentation, performed in the AI-Lab of Information and Communication Systems Engineering department of the University of the Aegean. Since the historical document images often suffer degradation problems, such as stains, background of big variations and uneven illumination, ink seepage, etc. the necessity of those tasks for the processing of historical documents will be shown. In the last years, since many systems have reached the best possible performance that a computer system can have the interaction with the user intend to improve them further. The importance of human feedback over document image processing tasks is going to be demonstrated. Due to the fact that the historical documents often are degraded and suffer from noise, and imperfect structure classic OCR techniques fail to issue this problem, word spotting is the only solution. It will be presented our work in word-spotting and our future plans.