Somatosensory dynamic gesture recognition method fusing image and physiological signal double channels

Abstract

The invention discloses a somatosensory dynamic gesture recognition method fusing two channels of an image and a physiological signal. The method comprises the following specific steps: 1, collectinga dynamic gesture physiological signal and an image sample; 2, carrying out noise reduction and frequency reduction processing on the physiological signals; 3, extracting image features by an image channel convolutional neural network; 4, establishing a multi-channel somatosensory gesture LSTM recognition model; 5, performing real-time testing based on the optimal somatosensory dynamic gesture recognition model. The fusion recognition framework is constructed based on the deep learning theory, the somatosensory dynamic gesture recognition model is constructed through the convolutional neural network and the long-short-term memory network, the recognition precision is high, the real-time performance is good, and the recognition robustness is good.

Description

Technical field [0001] The invention relates to the field of dynamic gesture recognition, in particular to a somatosensory dynamic gesture recognition method that combines image and physiological signal dual channels. Background technique [0002] With the development of economy and technology, people have higher and higher requirements for the quality of life. The somatosensory interaction mode satisfies the characteristics of nature, friendliness and humanity. It can obtain and process information at any time, any place, and in any way. It highlights the core idea of ​​"people-centered" and is useful for the development of universal intelligent computing and human-machine intelligence. Interactive technologies are of great significance, and thus directly play an important role in virtual reality, sports rehabilitation, smart home and other fields. Among them, somatosensory dynamic gesture recognition is one of the key modules of somatosensory interaction technology, and it is ...

AI Technical Summary

Problems solved by technology

Among them, the two most mainstream methods today are somatosensory gesture recognition technology based on images and physiological signals, but the image-based method has a limited observation range and is easily affected by various factors such as lighting conditions and occlusions, while the measurement noise based on physiological signals is limited. It is very sensitive and the individual signals vary greatly. It can be seen that these two mainstream methods cannot fully meet the needs of somatosensory dynamic gesture recognition. For this, a fusion somatosensory gesture recognition solution is needed, combining these two methods from the perspective of multi-sensor fusion. Get up, complement each other

Images