Estimating the pen trajectories of static signatures using hidden Markov models

Abstract

Static signatures originate as handwritten images on documents and by definition do not contain any dynamic information. This lack of information makes static signature verification systems significantly less reliable than their dynamic counterparts. This study involves extracting dynamic information from static images, specifically the pen trajectory while the signature was created. We assume that a dynamic version of the static image is available (typically obtained during an earlier registration process). We then derive a hidden Markov model from the static image and match it to the dynamic version of the image. This match results in the estimated pen trajectory of the static image.

Automatic Signature Verification: The State of the Art

“…In particular, pressure information, which can be registered with respect to various velocity bands, has been exploited for signature verification in order to take advantage of interfeature dependencies [154]. Furthermore, direction of pen movement [363], [366] and pen inclination [130], [151], [238] have also been successfully considered to improve the performance in online signature verification, whereas pen trajectory functions have been extracted from static signatures, in order to exploit the potential of dynamic information for offline signature verification as well [226]. Recent studies also demonstrate that signature verification can be successfully performed by means of "motif" series, which are characteristic subsequences extracted from function features [109].…”

Data Embedding into Characters

“…This estimation is so-called the stroke recovery problem [19]- [21] and a famous difficult inverse problem. However, if the pattern is not very complex, we still can recover the order correctly by using a sophisticated stroke recovery method like [20].…”

On the possibility of instance-based stroke recovery

“…Therefore this elegant method cannot deal with " " and " ". Most of other existing methods, such as [2], [3], [4], have the same difficulties because they also formulated the stroke recovery problem as a trace problem.…”