Background: Reconstructing the trajectory from the static image of handwritten ink traces is useful in many practical applications envisaging handwriting analysis and recognition from off-line data, as it allows to use methods, algorithms and tools that deal with on-line data, achieving better results than those achieved on off-line data.
Methods: In this work we addressed the trajectory recovery by proposing an approach inspired by the processes involved in human learning of motor skills to perform voluntary and complex movements. As humans learn motor skills by a trial-and-error process driven by the performance and the consumption of metabolic energy, our approach generates a trajectory and estimates the consumption of metabolic energy needed to execute it, and in case it is deemed too energy demanding, a new one is generated and its energy consumption is evaluated. Eventually the one corresponding to the minimum energy consumption among the extant ones is selected as the actual one.
Results and Conclusions: The effectiveness of the proposed approach has been quantitatively and extensively evaluated on a large and publicly available dataset, containing multi-stroke words. The experimental results show that our approach outperforms the existing ones in terms of Root Mean Square Error and Dynamic Time Warping distance between the recovered trajectories and the actual ones. Furthermore, an on-line recognition system provided with the trajectory recovered from off-line samples showed an overall reduction of about 1% with respect to the recognition rate achieved by the system when provided with on-line data.