Implicit Palm Rejection Using Real-Time Hand Model Filters on Tablet Devices

Abstract

Most tablet devices usually suffer from the "palm rejection problem", where unintended multiple touches while writing often cause erroneous application behavior and unsightly imprints on the display. We designed a real-time implicit palm rejection algorithm based on hand model filters and touch characteristics. We focus our approach on accurately determining the context of the touch in real-time to provide users with an experience that is close to natural handwriting as possible. Our algorithm uses a model-based filtering method to define the palm rejection region and automatically adjust to palm-first or write-first scenarios. Our implementation can correctly filter 99% of stylus touches with a low errant touch rate of 0.87%. In summary, our palm rejection algorithm is shown to be comparable or better than most currently available note-taking applications.