%0 Conference Proceedings
%4 dpi.inpe.br/vagner/1999/11.26.16.43
%2 sid.inpe.br/banon/2001/03.30.15.57.38
%@doi 10.1109/SIBGRA.1999.805725
%T Segmentation-free printed character recognition by relaxed nearest neighbor learning of windowed operator
%D 1999
%A Kim, Hae Yong,
%E Stolfi, Jorge,
%E Tozzi, Clésio Luis,
%B Brazilian Symposium on Computer Graphics and Image Processing, 12 (SIBGRAPI)
%C Campinas, SP, Brazil
%8 17-20 Oct. 1999
%I IEEE Computer Society
%J Los Alamitos
%P 195-204
%S Proceedings
%1 SBC - Brazilian Computer Society and UNICAMP - University of Campinas
%K optical, recognition, windowed operation.
%X The segmentation is considered by many researchers as the key technology for a reliable optical character recognition (OCR) system. To accomplish a sound segmentation, many alternative techniques have been recently proposed. This paper presents a new technique to recognize characters without an explicit segmentation. It is based on the automatic construction of windoed operator by relaxed nearest neighbor learning. It has been implemented, tested and yielded excellent recognition accuracy and computational performance.
%3 001-70833333.pdf
%O The conference was held in Campinas, SP, Brazil, from October 17 to 20.