Duplicate document detection

摘要: In document image filing applications it is important to be able recognize whether a particular has already been entered into the system either as an individual or inclusion in another document.Document images could matched on basis of layout contents.However, matching may not effective when style strictly controlled. We develop 'handle' which stored along with image. The handle simply character shape coded representation after figures and tables have ben removed. Character coding method identifying members one small number classes. This process computationally inexpensive tolerant differing generations photocopying, skew scanner characteristics. When new system, its computed compared against al extant handles using normalized Levenshtein metric. demonstrate ability detect duplicate documents comprising single multiple pages.