Generating Handwriting via Decoupled Style Descriptors
This work addresses the challenge of generating flexible and high-quality handwriting for applications like digital writing or personalization, though it is incremental as it builds on existing VRNN approaches.
The paper tackled the problem of representing handwriting stroke styles by decoupling character- and writer-level styles, resulting in generated handwriting preferred over a state-of-the-art baseline 88% of the time and achieving 89.38% accuracy in writer identification from a single sample word.
Representing a space of handwriting stroke styles includes the challenge of representing both the style of each character and the overall style of the human writer. Existing VRNN approaches to representing handwriting often do not distinguish between these different style components, which can reduce model capability. Instead, we introduce the Decoupled Style Descriptor (DSD) model for handwriting, which factors both character- and writer-level styles and allows our model to represent an overall greater space of styles. This approach also increases flexibility: given a few examples, we can generate handwriting in new writer styles, and also now generate handwriting of new characters across writer styles. In experiments, our generated results were preferred over a state of the art baseline method 88% of the time, and in a writer identification task on 20 held-out writers, our DSDs achieved 89.38% accuracy from a single sample word. Overall, DSDs allows us to improve both the quality and flexibility over existing handwriting stroke generation approaches.