Audio Based Handwriting Input for Tiny Mobile Devices. Yu, T., Jin, H., & Nahrstedt, K. In Proceedings - IEEE 1st Conference on Multimedia Information Processing and Retrieval, MIPR 2018, 2018. IEEE.
Audio Based Handwriting Input for Tiny Mobile Devices [pdf]Paper  Audio Based Handwriting Input for Tiny Mobile Devices [link]Website  abstract   bibtex   
The popularization of tiny mobile devices has raised the problem that it is hard to efficiently input messages via tiny keyboards or touch screens. In this paper, we present TableWrite, an audio-based handwriting input scheme, which allows users to input words to mobile devices by writing on tables with fingers. The key feature is that, once trained by a user, TableWrite does not require any retraining phase before each use. To reduce the impacts of audio signal's multipath propagation, we design multiple features that maintain consistency even when writing positions keep changing. We apply machine learning and gesture tracking techniques to further improve the accuracy of handwriting recognition. Our prototype system's experimental results show that the average accuracy of word recognition is around 90%-95% in lab environments, which validates the effectiveness of TableWrite.
@inProceedings{
 title = {Audio Based Handwriting Input for Tiny Mobile Devices},
 type = {inProceedings},
 year = {2018},
 identifiers = {[object Object]},
 keywords = {Audio Signals,Handwriting Recognition},
 websites = {https://ieeexplore.ieee.org/document/8396986},
 publisher = {IEEE},
 city = {Miami, FL},
 id = {84b7286f-a3c1-3801-852a-e0bbd0423a88},
 created = {2019-01-29T17:30:16.423Z},
 file_attached = {true},
 profile_id = {9e0c8f3c-9cd7-3bea-bf08-594485e332bb},
 group_id = {c1e9e98b-e33c-3718-9343-9cc7f3e51d80},
 last_modified = {2019-02-28T14:27:39.768Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 citation_key = {Yu2018},
 private_publication = {false},
 abstract = {The popularization of tiny mobile devices has raised the problem that it is hard to efficiently input messages via tiny keyboards or touch screens. In this paper, we present TableWrite, an audio-based handwriting input scheme, which allows users to input words to mobile devices by writing on tables with fingers. The key feature is that, once trained by a user, TableWrite does not require any retraining phase before each use. To reduce the impacts of audio signal's multipath propagation, we design multiple features that maintain consistency even when writing positions keep changing. We apply machine learning and gesture tracking techniques to further improve the accuracy of handwriting recognition. Our prototype system's experimental results show that the average accuracy of word recognition is around 90%-95% in lab environments, which validates the effectiveness of TableWrite.},
 bibtype = {inProceedings},
 author = {Yu, Tuo and Jin, Haiming and Nahrstedt, Klara},
 booktitle = {Proceedings - IEEE 1st Conference on Multimedia Information Processing and Retrieval, MIPR 2018}
}
Downloads: 0