Developped a non parametric handwriting synthesis algorithm
Winner of the prize (sponsored by the BBC) for best project of the year.
Created a novel method to synthesize a person’s handwriting style. The approach taken was a data-driven non-parametric one that was inspired by ideas… · More proposed in non-parametric texture synthesis algorithms (e.g. Image Quilting). The end result was an adaptable algorithm that is able to cope with both fully joined up and disjoint (and in-between) handwriting styles. A user study was conducted to determine how successful the algorithm was based on a number of different metrics. The majority of users agreed that the algorithm: -Produced output that was difficult to distinguish from samples of the original writers hand-writing -Produced realistic (i.e. human-looking) handwriting -Produced output that reflected the original authors personal style of writing