Multi-device cooperative gait sensing and identity recognition method based on ultrasonic waves

Abstract

The walking direction is calculated according to the walking path; the method comprises the following steps: selecting a speed collected by a transceiver close to a user according to a principle of proximity to synthesize a real walking speed of the user, respectively extracting different percentage speeds to extract fine-grained gait characteristics, and establishing a binary classification identity recognition model for each user through a machine learning method SVM to recognize identities of different people. According to the multi-device cooperative gait sensing and identity recognition method based on ultrasonic waves, two pairs of sensing devices are deployed to weaken walking path constraints; doppler frequency spectrum information captured by different devices is combined, the real-time position of the user is tracked, the walking direction is judged, the step speed is calculated, speed synthesis is conducted according to the triangular relation between the user and the positions of the two sensing devices, the real speed of the user is obtained, and fine-grained gait characteristics are extracted to recognize the identity of the user.

Description

technical field [0001] The invention relates to the field of human behavior perception based on ultrasonic signals, in particular to a user walking path tracking and fine-grained gait perception method in an indoor environment. Background technique [0002] Identification technology is playing an increasingly important role in our daily life. The mainstream identification technology in the existing market is mainly based on technologies such as computer vision images, based on biological characteristics such as face, fingerprint and iris. Users need to be close to the sensing device to collect data, which is sensitive to factors such as light and distance, so it cannot support long-distance identification and there is a risk of privacy leakage, and it is difficult to meet the identification needs of special environments such as insufficient lighting. At present, there are many new identification methods based on wireless signals (Wi-Fi, radar, etc.). It has inherent advanta...

AI Technical Summary

Problems solved by technology

[0002]Identification technology is playing an increasingly important role in our daily life. The mainstream identification technology in the existing market is mainly based on technologies such as computer vision images. Biometric features such as face, fingerprint, and iris are used as the basis for identification. During identification, the user needs to be close to the sensor device to collect data, which is sensitive to factors such as light and distance. Therefore, it cannot support long-distance identification and there is a risk of privacy leakage. It is difficult to meet the needs of insufficient light. Identification requirements in special environments such as

At present, there are many new identification methods based on wireless signals (Wi-Fi, radar, etc.). It has inherent advantages such as slow transmission speed, and can achieve good results in the field of identification. In the article "AcousticID: Gait based Human Identification Using Acoustic Signal" published by UbiComp in 2019, we proposed a gait perception and recognition based on ultrasonic signals. The identity recognition method can be used as a solution for identity control systems such as access control, but the main disadvantage of this method is that only a single pair of audio transceiver devices is used, and the user is required to walk a sufficient distance (2-3m) in a straight line in an open area. Indoor environments such as offices provide identification services, so it is necessary to combine multiple transceiver devices to study indoor identification methods

Images