SIGNAL CONDITIONING AND PERFORMANCE OPTIMIZATION OF A FLEX SENSOR-BASED SIGN LANGUAGE RECOGNITION SYSTEM
Keywords:
Signal conditioning, flex sensors, sign language recognition, digital filtering, hysteresis control, sensor calibration, performance optimization, embedded systemsAbstract
ABSTRACT
Flex sensor-based sign language recognition systems offer a low-cost and portable alternative to vision-based gesture recognition frameworks. However, inherent nonlinear sensor behavior, electrical noise, analog-to-digital conversion fluctuations, and unstable threshold logic can degrade system stability and responsiveness. This study presents a structured signal conditioning and performance optimization framework applied to a previously developed flex sensor-based sign language recognition system. The proposed improvements include microcontroller-level digital filtering using moving average, low-pass, and median filters; hysteresis-based threshold control; user-specific calibration and normalization; mathematical modeling of sensor transfer characteristics; quantitative stability evaluation using standard deviation and signal-to-noise ratio; and response time analysis. Experimental results demonstrate a reduction in signal fluctuation from ±18 ADC units to ±3 ADC units and latency improvement from 480 ms to 360 ms. The framework enhances robustness without increasing hardware complexity and aligns with established principles in control systems and measurement engineering.
