Comparative Evaluation of Analog and Digital Filtering for Non-Contact Capacitive Hand Kinematics Sensing

Non-contact capacitive sensing enables low cost and unobtrusive measurement of hand kinematics for human computer interaction and rehabilitation applications. However, capacitive sensors are highly susceptible to environmental electromagnetic interference, making effective filtering essential for reliable signal extraction. This work presents a systematic comparison of analog, digital and cascaded filtering strategies for non-contact capacitive hand kinematics sensing. A fourth order Butterworth low-pass filter with a cutoff frequency of 25 hertz was designed as an anti-aliasing front end and evaluated against fourth order digital infinite impulse response and finite impulse response filters. Performance was assessed using signal to noise ratio, latency and power consumption metrics on controlled sensor-like signals validated with electric potential sensor measurements. Results show that digital filtering achieved a standalone signal to noise ratio of 26 decibels while analog filtering achieved 17 decibels with negligible latency and a 91 percent reduction in processing power. The cascaded architecture achieved the highest signal quality with a signal to noise ratio of 29 decibels while preserving real time performance and low power operation. These results provide practical design guidance for robust capacitive hand kinematics systems.