Title :
Neural network system for continuous hand-written words recognition
Author :
Kussul, Ernst M. ; Kasatkina, Lora M.
Author_Institution :
Centro de Instrum., Univ. Nacional Autonoma de Mexico, Mexico City, Mexico
Abstract :
A method of continuous hand-written word recognition is proposed. The method is based on segmentation of the word onto triplets. Each triplet contains 3 letters. Two subsequent triplets have 2 common letters. Such overlapping gives powerful means for correction of recognised triplets and points of word segmentation. The proposed method could be used for creation the systems of automatic input of documents from handwritten archives into computers
Keywords :
content-addressable storage; document image processing; handwritten character recognition; image coding; image segmentation; learning (artificial intelligence); neural nets; continuous hand-written words recognition; handwritten archives; triplets; word segmentation; Assembly; Cleaning; Cybernetics; Feature extraction; Histograms; Image recognition; Image segmentation; Instruments; Neural networks; Text recognition;
Conference_Titel :
Neural Networks, 1999. IJCNN '99. International Joint Conference on
Conference_Location :
Washington, DC
Print_ISBN :
0-7803-5529-6
DOI :
10.1109/IJCNN.1999.833536
