Huge amounts of historical data are currently "buried" behind thousands of terabytes of handwritten text images. Automated methods are needed to make such important textual content accessible to scholars and users in general. OCR transcription technology is generally useless for this kind of documents because it is practically impossible to reliably segment the images of interest into individual characters or even into individual words.  This is why holistic Handwritten Text Recognition (HTR) approaches have emerged in the last two decades.

This lecture will present state-of-the-art HTR technologies for the transcription of handwritten text images.  Both fully automatic and semi-automatic, computer-assited HTR will be presented.  Finally, HTR applications where hand-held devices can be used for (camera-based) image acquisition will be outlined, along with multimodal interactive techniques which can be used to take advantage of the multiple input modalities of these devices.

Empirical results with real, large image collections will be reported and live demonstrations with these collections will be presented.  Moreover, the attendants will have the opportunity of a hands-on experience of using these advanced handwritten image transcription systems.