Date: Monday, September 14, 2009, 11:30 am, SB 239
Title: Enhancing Text Readability in Damaged Documents
Speaker: Dr. Gideon Frieder
A. James Clark Professor of Engineering and Applied Science
The George Washington University
Documents can be damaged for various reasons — deliberate distortion, aging, natural cause such as floods, etc. In the preliminary work reported herein, we present some results of processes that enhance the visibility of lines, therefore the readability of text in such documents. No attempt is made to interpret the contents — rather, the work intends to aid an analyst that will eventually process the information that is now easier to see and acquire.
Our developed tool provides different methods of enhancement, and the analyst can choose the method and set the optimal parameters. The results that we have are encouraging, however, the processing is done on every page separately and parameter setting to achieve optimal enhancement is difficult. The key obstacle preventing automation is the absence of a mathematical definition of readability and a figure of merit whose value will guide the automatic selection of optimal parameters. We shall present our results, and hope that it will motivate the listeners to help in finding such figure of merit.
Data used in this presentation consists of the diaries of HaRav Dr. Avraham Abba Frieder. These diaries consist of more then 800 pages written in different scripts, both typed and handwritten, in different languages, in different states of readability; with intermix of text, pictures, tables, corrections etc. The diaries are now on a permanent loan to the archives of Yad Va Shem, and are found in the USHMM archives and at http://www.ir.iit.edu/collections.
Biography: Dr. Gideon Frieder is the A. James Clark Professor of Engineering and Applied Science and Professor of Statistics at The George Washington University where he previously also served as Dean of the School of Engineering and Applied Science.