An Optimization of IMU Sensors-Based Approach for Malaysian Sign Language Recognition

Abstract

Malaysian Sign Language (MSL) has been the main communication medium for Malaysian deaf people but only a small portion of ordinary people can comprehend this MSL. A common approach to convey the meaning of sign language is by using sign language recognition (SLR) method. A video-based SLR approach has been researched extensively but still heavily vulnerable to environment error and suffer from portability issue. Conversely, a sensor-based SLR solution can mitigate these problems at the cost of having bulky design due to the number and placement of the sensors required to achieve sufficient accuracy. Therefore, in this paper, we investigated the optimal number and placement of the inertial measurement unit (IMU) sensors required to perform an accurate conversion. Experimental results demonstrated our proposed optimization process is capable of achieving more than 98% of recognition accuracy. This could be a benchmark for a less bulky and leaner design of sensor-based hand glove as well as overcoming the usability and portability issues.