Hidden Markov Model Based Word Recognition and Its Application to Legal Amount Reading on French Checks

Recognition-directed recovering of temporal information from handwriting images

“…Many factors differentiate these two domains: application areas, digitization process, data formats used and richness of information regarding the handwriting. The most prominent applications for offline recognition systems are bank check processing (Knerr et al, 1998), form processing (Cracknell and Downton, 1998) and postal address reading (Srihari and Keubert, 1997). In these situations, an already existing document is converted into an image using a camera or a scanner.…”

Off-line handwritten word recognition using multi-stream hidden Markov models

“…While many of such systems are already providing good performance for several applications like OCR, Bank checks Readers, Forms readers, etc. (Srihari 2000) (Knerr et al, 1998) (D'Amato and Kuebert, 2000) (Gorski et al, 1999), the enhancement of performance is still required so as to cope with a wider range of document reading applications.…”

Recognition of cursive Roman handwriting: past, present and future

“…The features describe structural properties, such as ascenders, descenders, loops or cusps, or they are derived from the greylevel distribution of the pixels in the window. Examples of HMM-based word recognizers can be found in [7,27,34,57,64,68,81,127]. As an extension of pure HMM-based recognition, some authors have proposed to use HMMs together with neural networks in hybrid systems [65,97,105] A special case of cursive words is digit string.…”