Combined Dynamic Time Warping with Multiple Sensors for 3D Gesture Recognition.

Area of Science:

• Cyber-physical Systems
• Human-Computer Interaction
• Sensor Fusion

Background:

• Cyber-physical systems (CPS) leverage user movement analysis for broader applications.
• Three-dimensional (3D) gesture recognition typically requires multiple sensors for natural gesture identification.
• Existing methods struggle with real-time data capture and integration from independent sensors.

Purpose of the Study:

• To propose a robust 3D gesture recognition method using combined multi-sensor information.
• To address challenges in real-time data capture and fusion for gesture recognition.
• To enhance the performance of 3D gesture recognition regardless of user position or movement.

Main Methods:

• Development of a multi-sensor fusion approach for 3D gesture recognition.
• Implementation of viewpoint-weighted and/or motion-weighted values for robust recognition.
• Utilizing viewpoint-weighted dynamic time warping with multiple sensors to mitigate errors.

Main Results:

• The proposed method demonstrates robust gesture recognition across different user locations and movement directions.
• Viewpoint-weighted dynamic time warping effectively prevents joint measurement errors and sensor noise.
• Enhanced recognition performance achieved through effective comparison of multiple joint sequences.

Conclusions:

• The multi-sensor fusion method offers a significant improvement in 3D gesture recognition accuracy and robustness.
• This approach simplifies data capture and combination for real-time applications.
• The technique holds promise for advancing human-computer interaction within cyber-physical systems.