Building on our 10-year-experience with script recognition systems, a new reader generation was designed. Previously, only single handwritten words were compared against a dictionary. Now a short text is modeled and processed as a whole. The system does not proceed in a linear fashion anymore, but uses feedback between image processing, character recognition and syntactic alignment. The system features a new recognition kernel that works without prior segmentation and is based on geometric form descriptors. © Springer-Verlag Berlin Heidelberg 2004.
CITATION STYLE
Boldt, M., & Asp, C. (2004). Recognition of short handwritten texts. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2956, 91–102. https://doi.org/10.1007/978-3-540-24642-8_6
Mendeley helps you to discover research relevant for your work.