Machine Learning Method-Based Static Infrared Gesture Recognition System

Infrared sensor arrays are extensively utilized in applications such as human activity monitoring and gesture recognition systems. The high-density configuration of these sensor arrays ensures a reliable and consistent platform for uncontact signal detection. This letter presents a machine-learning approach for infrared gesture recognition. The infrared sensor array employed in this study comprises 16 readout channels, each equipped with a corresponding voltage amplification unit. The sensors are spaced 3 cm apart, while the voltage amplifier consumes 4.77 W. A back propagation (BP) neural network is used to process and classify gesture data. To improve accuracy, the weights of the BP neural network are initialized by using a genetic algorithm. Experiments show that infrared gesture signals can be classified effectively by the proposed modified BP neural network, which achieves an accuracy of over 96% on the test set.