Kooka and KADMOS integration ============================ This file describes how to make Kooka working with the KADMOS OCR/ICR engine. KAMDOS is commercial OCR/ICR software component of the company reRecognition GmbH Hafenstr. 50B 8280 Kreuzlingen Switzerland Tel.: +41 71 6780000 Fax: +41 71 6780099 www.reRecognition.com and Kooka can be build using the linux version of the component in order to achive very good ocr results. Please contact re Recognition directly if you like to obtain or test KADMOS. Note that if you are linking against KADMOS, you loose the permission to use the GPL Qt version, so you need a commercial Qt version as well. Configuration ------------- As Kooka does not require KADMOS, it is neccessary to configure Kooka to use the KADMOS library. This could be done by calling the configure script of Kooka's source distribution with following parameter: ./configure --with-kadmos-dir=/where/kadmos.h/resides/ The configure script tries to locate the file kadmos.h in the directory that was specified. The KADMOS library is expected in the same directory. Build the source after configuring with KADMOS. Kooka enables the code to work with KADMOS in the source and links the library. Installation ------------ KADMOS is linked statically and thus there is no need for special installation of the KADMOS library. Installation of the Classifier Files: KADMOS needs classifier files for the ocr process which come with the KADMOS developer's toolkit. The classifier files need to be installed in the KDE application data directory for Kooka in a subdirectory named classifiers. If your KDE installation goes to /opt/trinity/, this is for example /opt/trinity/share/apps/kooka/classifiers. Kooka picks the available classifiers up automatically. The classifiers are named in the following way: [fontkind][country/region].rec, where fontkinds are ttf -> machine print font hand -> handprint (isolated) norm -> OCR norm font For example the following classifier names are used: ttfus.rec US machineprint classifiers handus.rec US handwriting classifiers norm.rec Special OCR character sets, not localized ---------------------------------------------------------------------- Klaas Freitag $Id$