Advancement in smartphones has facilitated the investigation of new modalities of human-machine interaction, including communication through touch, voice, and gestures. In-depth, the researchers examined the problem of recognizing distinct gestures… Click to show full abstract
Advancement in smartphones has facilitated the investigation of new modalities of human-machine interaction, including communication through touch, voice, and gestures. In-depth, the researchers examined the problem of recognizing distinct gestures (surface, hand, and motion). However, the gesture recognition algorithm pitches discontinuity while the user performs the subsequent continuous gesture. The discontinuity may occur due to the selection of a delimiter to differentiate between successive motions or the employment of a complex algorithm to boost the accuracy of gesture detection, which takes significant time to recognize the gesture before a user may enter the next gesture. Further, gesture recognition based on template matching, machine learning models, and neural networks requires a lot of storage space, processing resources, or both, which are resource-intensive for smartphones. This research proposes a novel Axis-Point Continuous Motion Gesture (APCMG) recognition algorithm that uses accelerometer sensor data to recognizes continuous motion gestures in real time. The algorithm has low computational complexity and easily implemented on resource-constrained devices with minimal computing cost, memory, and energy. The prime objective of the APCMG is to find the start and end of a gesture from a continuous stream of accelerometer sensor data and recognize the gesture in real-time. To demonstrate the APCMG efficacy, the experimental simulation of the Android application for dialing a phone number is considered. The App acknowledges 12 continuous gestures corresponding to 0 to 9 number, delete, and calls termination. The experimental simulations collected 7500 gestures samples from the 25 volunteers. The algorithm efficiently recognizes isolated and continuous gestures with 95% and 94% accuracy, respectively. The proposed algorithm efficiently recognizes isolated and continuous gestures with minimal energy consumption.
               
Click one of the above tabs to view related content.