Malaysian sign language recognition framework based on sensory glove
The purpose of this study was to propose a low-cost and real-time recognition system using asensory glove, which has 17 sensors with 65 channels to capture static sign data of the Malaysiansign language (MSL). The study uses an experimental design. Five participants well-known MSLwere chosen to perf...
Saved in:
| Main Author: | |
|---|---|
| Format: | thesis |
| Language: | eng |
| Published: |
2019
|
| Subjects: | |
| Online Access: | https://ir.upsi.edu.my/detailsg.php?det=5224 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The purpose of this study was to propose a low-cost and real-time recognition system using asensory glove, which has 17 sensors with 65 channels to capture static sign data of the Malaysiansign language (MSL). The study uses an experimental design. Five participants well-known MSLwere chosen to perform 75 gestures throughout wear sensory glove. This research was carriedout in six phases as follows: Phase I involved a review of literature via a systematic reviewapproach to identify the relevant set of articles that helped formulate the research questions.Phase II focused on the analysis of hand anatomy, hand kinematic, and hand gestures to helpunderstand the nature of MSL and to define the glove requirements. In Phase III, DataGlove wasdesigned and developed based on the glove requirements to help optimize the best functions of theglove. Phase IV involved the pre-processing, feature extraction, and classification of the datacollected from the proposed DataGlove and identified gestures of MSL. A new vision and sensor-basedMSL datasets were collected in Phase V. Phase VI focused on the evaluation and validation processacross different development stages. The error rate was used to check system performance. Also, a3D printed humanoid arm was used to validate the sensor mounted on the glove. The results of dataanalysis showed 37 common patterns with similar hand gestures in MSL. Furthermore, thedesign of DataGlove based on MSL analysis was effective in capturing a wide range of gestures witha recognition accuracy of 99%, 96%, and 93.4% for numbers, alphabet letters, and words,respectively. In conclusion, the research findings suggest that 37 group's gestures of MSLcan increase the recognition accuracy of MSL hand gestures to bridge the gap between peoplewith hearing impairments and ordinary people. For future research, a more comprehensive analysis of the MSL recognition system isrecommended. |
|---|