Audio Doppler characteristic quantification-based gesture identification method

Abstract

The invention relates to an audio Doppler characteristic quantification-based gesture identification method and belongs to the field of man-machine interaction. By utilizing a general audio playing and receiving device, a Doppler effect of an audio signal is extracted to realize gesture identification. The method comprises the steps of firstly playing the audio signal with a certain frequency by a loudspeaker; capturing the audio signal reflected by a human body through a microphone; forming an eigenvector set by extracting a fine spectral structure after Doppler frequency shift; and realizing gesture identification classification through gesture primitive segmentation and an HMM. The gesture identification precision is extremely high and the gesture types are rich; the difficulty of incapability of accurately identifying gestures with same absolute distance paths in a previous audio Doppler gesture identification technology is overcome; the calculation is simple; the practicality is high; and the method can be widely applied to the fields of somatosensory interaction, virtual reality, smart home and the like.

Description

technical field [0001] The invention belongs to the technical field of human-computer interaction, and in particular relates to a bare-hand gesture recognition method based on audio Doppler feature quantization. Background technique [0002] As a simple and natural way of interaction, gestures are more in line with people's communication habits. Gesture recognition technology is an important way for natural human-computer interaction. Gesture recognition can be realized based on vision, inertial sensors of wearable devices, and Doppler effect of fluctuating signals. The sensitivity of visual recognition methods to light and the constraints of wearable devices on the use process all limit its application range. Gesture recognition based on the Doppler effect of fluctuating signals can well overcome the shortcomings of the above two methods. For this reason, there have been a variety of gesture recognition methods based on the Doppler effect, such as published patents: CN10379...

AI Technical Summary

Problems solved by technology

However, there are also some deficiencies in these methods or devices, which do not consider the audio Doppler effect in gesture recognition applications, the ambiguity of frequency shift features and energy features, resulting in low gesture recognition accuracy and limited gesture types

Images