Comparing Normalization and Adaptation Techniques for On-Line Handwriting Recognition

In this paper a writer-independent on-line handwriting recognition system is described comparing the influence of handwriting normalization and adaptation techniques on the recognition pe@ormance. Our Hidden Markov Model (HMM) -based recognition system for unconstrained German script can be adapted to the writing style of a new writer using d#erent adaptation techniques whereas the impact of preprocessing to normalize the pen-trajectory is examined. The performance of the resulting writerdependent system increases significantly, even if only a few words are availablefor adaptation. So this approach is also applicablefor on-line systems in hand-held computers such as PDAs. In addition, the developed normalization techniques are helpful to improve completely writer independent systems. This paper presents the performance comparison of three d$ferent adaptation techniques either in a supervised or an unsupervised mode, in combination with appmpriate normalization methods, with the availability of d#erent amounts of adaptation data ranging from only 6 words up to 100 words per writer.