Static and Dynamic Synthesis of Bengali and Devanagari Signatures

Abstract

Developing an automatic signature verification system is challenging and demands a large number of training samples. This is why synthetic handwriting generation is an emerging topic in document image analysis. Some handwriting synthesizers use the motor equivalence model, the well-established hypothesis from neuroscience, which analyses how a human being accomplishes movement. Specifically, a motor equivalence model divides human actions into two steps: 1) the effector independent step at cognitive level and 2) the effector dependent step at motor level. In fact, recent work reports the successful application to Western scripts of a handwriting synthesizer, based on this theory. This paper aims to adapt this scheme for the generation of synthetic signatures in two Indic scripts, Bengali (Bangla), and Devanagari (Hindi). For this purpose, we use two different online and offline databases for both Bengali and Devanagari signatures. This paper reports an effective synthesizer for static and dynamic signatures written in Devanagari or Bengali scripts. We obtain promising results with artificially generated signatures in terms of appearance and performance when we compare the results with those for real signatures.

A Perspective Analysis of Handwritten Signature Technology

“…However, other scripts have been also considered in the literature. Although they are not as large as the Western corpuses, we can find databases in Malaysian [3], Arabic [6], Bengali [47,74], Devanagari [74], Persian [98], Chinese [151], and Japanese [161]. It is worth pointing out that these databases are typically collected in the offline mode.…”

“…Acquired. To capture simultaneously online and offline signatures, a piece of paper is usually placed over the digitizing tablet (e.g., [74,75]). Thus, the same signature is collected for the two modalities without scale or rotation variation between them.…”

“…The system was evaluated on a Bengali dataset [192]. Another ASV based on texture features and SVM was applied to both Hindi and Bengali signatures in [51,74]. Thus, local binary patterns (LBPs) and uniform LPB strategies were used for parameterized Bengali and Hindi signatures in [192].…”

“…For online signatures, little work has been published. In [74], an ASV based on function features and DTW was used for OnOffSigBengali-75 and OnOffSigHindi-75 datasets. Only random forgery experiments were performed.…”

Fixed-sized representation learning from offline handwritten signatures of different sizes

“…We conducted experiments on four offline handwritten signature datasets: GPDS-960 [33], MCYT-75 [34], CEDAR [35] and Brazilian PUC-PR [36]; and two synthetic datasets, for Bengali and Devanagari scripts [25]. We used a subset of the GPDS-960 dataset for learn-ing feature representations, using the different architectures and training methods described in this article.…”