Gait camouflage method for resisting biological recognition

Abstract

The invention provides a gait camouflage method for resisting biological recognition. The gait camouflage method comprises the following steps: wearing a gait camouflage device on a key part of a user; the method comprises the following steps: acquiring acceleration characteristics, angular velocity characteristics and magnetic field signal data of a user through a gait camouflage device, fusing the acceleration characteristics, the angular velocity characteristics and the magnetic field signal data to form a super vector, and performing time sequence coding to obtain a sample super vector of a current time slice; calculating the Hamming distance between the sample supervector and the reference supervector in the subspace, and taking the Hamming distance as a classification basis; optimizing an electrical stimulation intervention strategy based on the gait recognition model; through functional electrical stimulation regulation and control of specific lower limb muscles, joint motion angles are influenced, space-time and kinematics parameters of gaits are changed, and thus gait characteristics with a camouflage effect are realized. Electrical stimulation is applied to lower limb muscles according to different gait stages, so that human gait features are effectively changed, and the gait camouflage target is achieved; by analyzing gait parameter differences under different camera visual angles, an intervention mode of maximizing gait changes is selected, and gait features can be effectively camouflaged.

Description

Technical Field

[0001] The invention relates to the technical field of gait recognition countermeasure, in particular to a gait camouflage method for counteracting biological recognition. Background Art

[0002] Human walking is a complex behavior pattern with significant individual differences, and its characteristics can be used for identity recognition. At present, gait recognition, as an emerging biometric recognition technology, has unique application prospects in the field of intelligent video surveillance due to its advantages such as long-distance non-contact and difficulty in camouflage. However, in scenarios such as military reconnaissance, individual gait characteristics are difficult to disguise, which may lead to the identification of reconnaissance personnel, thereby affecting combat capabilities. On the other hand, the widespread application of gait recognition technology may also lead to the risk of personal privacy leakage. Therefore, effective countermeasures against gait recognition technology have become an important means of protecting privacy.

[0003] Existing gait recognition countermeasures can be divided into two categories: one is to interfere with the gait recognition model's recognition of the target by adding fake images or artificial noise to the surveillance video input. However, this method relies on the intrusion of the enemy's surveillance system and requires the interception of the video data stream. It is not effective for LAN surveillance systems with a high security level. Another type of technology is to confuse the gait recognition model by training individuals to change their gait patterns (such as walking with their feet turned inside out, limping, etc.), but this method is easy to attract the attention of monitoring personnel, resulting in the loss of camouflage effect. In addition, since the muscle activation and coordination patterns of individual gait are relatively fixed, deliberately changing gait is still difficult to effectively hide identity information. Summary of the invention

[0004] In view of the deficiencies of the prior art, the present invention provides a gait camouflage method for counteracting biometric identification. The present invention can effectively camouflage gait characteristics and significantly reduce the risk of an individual being identified and tracked by a monitoring system.

[0005] The technical solution of the present invention is: a gait camouflage method against biological recognition, comprising the following steps:

[0006] S1), wearing the gait camouflage device on the key parts of the user;

[0007] S2), collecting the user's three-axis acceleration characteristics, three-axis angular velocity characteristics and magnetic field signal data through the gait camouflage device;

[0008] S3), fusing the three-axis acceleration features, the three-axis angular velocity features and the magnetic field signal data to form a supervector and performing time series encoding to obtain a sample supervector of the current time segment;

[0009] S4), superimposing the motion sample supervectors of different time segments under the same label according to the gait stage label, generating a gait stage reference supervector, and calculating the Hamming distance between the sample supervector and the reference supervector in the subspace as a classification basis;

[0010] S5), optimizing the electrical stimulation intervention strategy based on the relationship between electrical stimulation parameters and gait characteristics; and collecting lower limb movement and spatiotemporal parameters through a multi-view video camera, selecting the electrical stimulation strategy with the greatest difference from natural gait parameters at different viewing angles, and integrating it into the system control logic;

[0011] S6) During walking, by regulating the function of specific lower limb muscles with electrical stimulation, the joint movement angle is affected, and the spatiotemporal and kinematic parameters of gait are changed, thereby achieving a gait feature with a camouflage effect.

[0012] Preferably, in step S1), the gait camouflage device includes an inertial sensor module, an electrical stimulation electrode module, and a main control unit, wherein the inertial sensor module is worn on the key part of the user's lower limbs; the electrical stimulation electrode module is attached to the user's tibialis anterior and gastrocnemius muscles; the inertial sensor module is used to collect three-axis acceleration characteristics, three-axis angular velocity characteristics and magnetic field signal data; and the collected data is sent to the main control unit in real time through the SPI protocol; the main control unit determines the gait stage through an embedded algorithm, and controls the electrical stimulation electrode module to generate electrical stimulation pulses with corresponding parameters.

[0013] Preferably, in step S1), the main control unit adopts a single-chip microcomputer with an ARM architecture; the electrical stimulation electrode module utilizes a rectangular wave generating circuit and a boosting circuit to generate rectangular wave electrical pulses with different pulse widths and frequencies, and outputs them to the tibialis anterior muscle and gastrocnemius muscle belly through the hydrogel electrode for functional electrical stimulation;

[0014] The inertial sensor module includes a three-axis acceleration sensor, a three-axis angular velocity sensor and a three-axis magnetometer.

[0015] Preferably, in step S1), the gait camouflage device also includes a power module, which includes a 3.7V rechargeable lithium battery, a boost circuit, a buck circuit, and multiple connection interfaces; the boost circuit is used to increase the 3.7V voltage to 5V and supply power to the main control unit and the electrical stimulation electrode module; the buck circuit steps down the 3.7V voltage to 3.3V to supply power to the inertial sensor module.

[0016] Preferably, step S3) specifically includes the following steps:

[0017] S31), assign a unique name supervector D to each N feature of the kinematic signal i, and discretize the real value at the current moment and map it into value supervectors V of different levels i ;

[0018] S32), name the super vector D i With value supervector V i A single feature supervector is generated by XOR operation, and all single feature supervectors are superimposed to form a synthetic supervector of the signal type;

[0019] S33), performing temporal coding on the unimodal supervector through a bit shift operation ρ;

[0020] S33), the synthetic supervectors of different signal types are fused through the concatenation operation Concat to obtain the sample supervector F corresponding to the current time segment; the fusion expression is:

[0021]

[0022] Where F is the sample supervector; T is the number of time slices; ρ is the bit shift operation; Concat is the concatenation operation; represents the i-th acceleration; represents the j-th angular velocity feature; represents the kth magnetic field signal type; n1, n2, n3 represent the number of acceleration features, angular velocity features, and magnetic field signal type features, respectively; represents the value supervector of the i-th acceleration feature at the t-th time segment, The supervector representing the value of the j-th angular velocity feature at the t-th time segment, A supervector representing the value of the k-th magnetic field signal feature at the t-th time segment; Represents the exclusive OR operation.

[0023] Preferably, step S4) specifically includes the following steps:

[0024] S41), according to the gait stage label, the numerical superposition of each element of the motion sample supervector G of different time segments under the same label is performed to obtain a new supervector with unchanged dimension;

[0025] S42), performing a sign function binarization on each element of the new supervector to generate a gait phase reference supervector M whose elements are all 1 or -1;

[0026] S43), perform XOR calculation on the sample supervector G and the gait stage reference vector M, obtain the Hamming distance between the sample supervector G and the gait stage reference vector M, and use it as the classification basis to predict the gait stage at the current moment, that is:

[0027]

[0028] Where, d H (G,M) represents the Hamming distance between G and M, g i and m i are the elements of G and M at the ith position, symbol Represents the exclusive-or operation.

[0029] Preferably, in step S5), the functional electrical stimulation strategy with the largest difference from the natural gait parameters is collected by a multi-view video camera and selected to be incorporated into the control logic, which specifically includes the following steps:

[0030] S51), the user wears the gait camouflage device and walks, applying electrical stimulation of different positions, intensities and durations to the lower limb muscles, and establishing a projection relationship between the electrical stimulation parameters and various gait characteristics;

[0031] S52), collecting lower limb kinematics and spatiotemporal parameters through multi-view video cameras, and quantitatively evaluating the effect of electrical stimulation on natural gait patterns;

[0032] S53), taking natural gait as a reference, extracting lower limb movement characteristics, constructing gait parameter characteristics corresponding to different types of gaits, adjusting the electrical stimulation strategy through a correspondence table between electrical stimulation parameters and gait characteristics, and simulating gait characteristics of different ages, genders or body shapes;

[0033] S54), comparing the motion models under different electrical stimulation strategies with the natural gait, selecting the strategy with the greatest difference from the natural gait, and integrating it into the system control as the optimal electrical stimulation scheme.

[0034] Preferably, in step S6), adjusting the gait motion parameters during walking to achieve camouflage specifically includes the following steps:

[0035] S61), applying electrical stimulation pulses to the tibialis anterior muscle and the gastrocnemius muscle through the electrical stimulation electrode module;

[0036] S62) During normal walking, electrical stimulation is applied to the gastrocnemius muscle during the swing phase and to the tibialis anterior muscle during the support phase to adjust the joint angle and change the gait parameters to achieve a camouflage effect.

[0037] The beneficial effects of the present invention are:

[0038] 1. The present invention uses a wearable inertial sensor to collect the motion signals of the lower limbs of the human body in real time during walking, and applies electrical stimulation to the lower limb muscles according to different gait stages, thereby effectively changing the gait characteristics of the human body and achieving the goal of gait camouflage;

[0039] 2. The present invention analyzes the differences in gait parameters under different camera viewing angles and selects an intervention method that maximizes gait changes, which can effectively disguise gait characteristics and significantly reduce the risk of individuals being identified and tracked by the monitoring system. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a schematic diagram of the framework of gait camouflage in an embodiment of the present invention;

[0041] Figure 2 It is a schematic diagram of the process of gait confrontation in an embodiment of the present invention;

[0042] Figure 3 : is a framework diagram of a gait camouflage device in an embodiment of the present invention;

[0043] Figure 4 It is a process framework diagram of gait stage division in an embodiment of the present invention;

[0044] Figure 5 This is a flowchart of the electrical stimulation regulation of the present invention;

[0045] Figure 6 This is a circuit framework diagram of the electrical stimulation electrode module of the present invention;

[0046] Figure 7 4 is a circuit framework diagram of the gait camouflage device of the present invention. DETAILED DESCRIPTION

[0047] The specific implementation of the present invention will be further described below in conjunction with the accompanying drawings:

[0048] like Figure 1 and 2 As shown, this embodiment provides a gait camouflage method for counteracting biometric identification, comprising the following steps:

[0049] S1), wearing the gait camouflage device on the key parts of the user;

[0050] like Figure 1-3 As shown, the gait camouflage device described in this embodiment includes an inertial sensor module, an electrical stimulation electrode module, a main control unit, and a power module, and the main control unit is connected to the inertial sensor module, the electrical stimulation electrode module, and the power module respectively. The inertial sensor module is worn on the key parts of the user's lower limbs; the electrical stimulation electrode module is attached to the user's tibialis anterior and gastrocnemius muscles; the inertial sensor module is used to collect three-axis acceleration characteristics, three-axis angular velocity characteristics, and magnetic field signal data; and the collected data is sent to the main control unit in real time through the SPI protocol; the main control unit determines the gait stage through the embedded algorithm, and controls the electrical stimulation electrode module to generate electrical stimulation pulses with corresponding parameters.

[0051] The main control unit adopts an ARM-based single-chip microcomputer of model STM32F407ZGT6. Figure 6 As shown, the electrical stimulation electrode module includes two pairs of electrodes, and the electrical stimulation electrode module utilizes a rectangular wave generating circuit and a boosting circuit to generate rectangular wave electrical pulses with different pulse widths and frequencies, which are output to the tibialis anterior and gastrocnemius muscle bellies through hydrogel electrodes for functional electrical stimulation.

[0052] The inertial sensor module includes a three-axis acceleration sensor, a three-axis angular velocity sensor and a three-axis magnetometer.

[0053] The gait camouflage device also includes a power module, such as Figure 7 As shown, the power module includes a 3.7V rechargeable lithium battery, a boost circuit, a buck circuit, and multiple connection interfaces; the boost circuit is used to increase the 3.7V voltage to 5V and power the main control unit and the electrical stimulation electrode module; the buck circuit steps down the 3.7V voltage to 3.3V to power the inertial sensor module.

[0054] S2), collecting the user's three-axis acceleration characteristics, three-axis angular velocity characteristics and magnetic field signal data through the gait camouflage device;

[0055] In this embodiment, three-axis acceleration characteristics, three-axis angular velocity characteristics and magnetic field signal data are collected by a three-axis acceleration sensor, a three-axis angular velocity sensor and a three-axis magnetometer respectively.

[0056] S3), the three-axis acceleration features, the three-axis angular velocity features and the magnetic field signal data are integrated to form a super vector and time-series encoded to obtain a sample super vector of the current time segment; Figure 4 As shown, the specific steps include:

[0057] S31), assign a unique name supervector D to each N feature of the kinematic signal i , and discretize the real value at the current moment and map it into value supervectors V of different levels i ; The kinematic signals in this embodiment are acceleration characteristics, three-axis angular velocity characteristics and magnetic field signals.

[0058] S32), name the super vector D i With value supervector V i A single feature supervector is generated by XOR operation, and all single feature supervectors are superimposed to form a synthetic supervector of the signal type;

[0059] S33), performing temporal coding on the single-mode synthetic supervector through a bit shift operation ρ;

[0060] S34), the encoded synthetic super vectors of different signal types are fused through the concatenation operation Concat to obtain the sample super vector F corresponding to the current time segment; the fusion expression is:

[0061]

[0062] Where F is the sample supervector; T is the number of time slices; ρ is the bit shift operation; Concat is the concatenation operation; represents the i-th acceleration; represents the j-th angular velocity feature; represents the kth magnetic field signal type; n1, n2, n3 represent the number of acceleration features, angular velocity features, and magnetic field signal type features, respectively; represents the value supervector of the i-th acceleration feature at the t-th time segment, The supervector representing the value of the j-th angular velocity feature at the t-th time segment, A supervector representing the value of the k-th magnetic field signal feature at the t-th time segment; Represents the exclusive OR operation.

[0063] S4), according to the gait stage label, superimpose the motion sample supervectors of different time segments under the same label to generate a gait stage reference supervector, and calculate the Hamming distance between the sample supervector and the reference supervector in the subspace as the classification basis; specifically comprising the following steps:

[0064] S41), according to the gait stage label, the numerical superposition of each element of the motion sample supervector G of different time segments under the same label is performed to obtain a new supervector with unchanged dimension;

[0065] S42), perform a sign function binarization on each element of the new supervector to generate a gait phase reference supervector M whose elements are all 1 or -1; in this embodiment, the gait phase labels include initial contact, mid-support, final support, swing phase, etc.

[0066] S43), calculate the Hamming distance between the sample supervector G and the gait stage reference vector M as the classification basis, and predict the gait stage at the current moment, that is:

[0067]

[0068] Where, d H (G,M) represents the Hamming distance between G and M, g i and m i are the elements of G and M at the ith position, symbol Represents the exclusive-or operation.

[0069] S5), optimize the electrical stimulation intervention strategy based on the relationship between electrical stimulation parameters and gait characteristics; collect lower limb movement and spatiotemporal parameters through multi-view video cameras, select the electrical stimulation strategy with the greatest difference from natural gait parameters at different viewing angles, and integrate it into the system control logic; Figure 5 As shown, the specific steps include:

[0070] S51), the user wears the gait camouflage device and walks, applies electrical stimulation of different positions, intensities and durations to the lower limb muscles, and establishes a projection relationship between the electrical stimulation parameters and various gait characteristics; as shown in Table 1:

[0071] Table 1 Projection relationship between various electrical stimulation parameters and various gait characteristics

[0072]

[0073] S52) Collect lower limb kinematics and spatiotemporal parameters through a multi-view video camera to quantitatively evaluate the effect of electrical stimulation on the natural gait pattern; wherein the spatiotemporal parameters include foot contact position, joint angles, limb movement speed, etc.

[0074] S53), taking natural gait as a reference, extracting lower limb movement characteristics (such as stride, step frequency, joint angle, etc.), constructing gait parameter characteristics corresponding to different types of gaits (such as different ages, genders or body shapes), adjusting the electrical stimulation strategy through the electrical stimulation parameter and gait characteristic correspondence table, and simulating gait characteristics under different ages, genders or body shapes;

[0075] S54), comparing the motion models under different electrical stimulation strategies with the natural gait, selecting the strategy with the greatest difference from the natural gait, and integrating it into the system control as the optimal electrical stimulation scheme.

[0076] S6) During walking, by regulating the function of specific lower limb muscles with electrical stimulation, the joint movement angle is affected, and the spatiotemporal and kinematic parameters of gait are changed, thereby achieving gait characteristics with camouflage effect, which specifically includes the following steps:

[0077] S61), applying electrical stimulation pulses to the tibialis anterior muscle and the gastrocnemius muscle through the electrical stimulation electrode module;

[0078] S62) During normal walking, electrical stimulation is applied to the gastrocnemius muscle during the swing phase and to the tibialis anterior muscle during the support phase to adjust the joint angle and change the gait parameters to achieve a camouflage effect.

[0079] The above embodiments and descriptions are only for illustrating the principles and best embodiments of the present invention. Without departing from the spirit and scope of the present invention, the present invention may be subject to various changes and improvements, all of which fall within the scope of the present invention to be protected.

Claims

1. A gait camouflage method for counteracting biometric identification, characterized in that: The following steps are involved: S1), wearing the gait camouflage device on the key parts of the user; S2), collecting the user's three-axis acceleration characteristics, three-axis angular velocity characteristics and magnetic field signal data through the gait camouflage device; S3), fusing the three-axis acceleration features, the three-axis angular velocity features and the magnetic field signal data to form a supervector and performing time series encoding to obtain a sample supervector of the current time segment; S4), superimposing the motion sample supervectors of different time segments under the same label according to the gait stage label, generating a gait stage reference supervector, and calculating the Hamming distance between the sample supervector and the reference supervector in the subspace as a classification basis; S5), optimizing the electrical stimulation intervention strategy based on the correspondence between electrical stimulation parameters and gait characteristics; collecting lower limb movement information and spatiotemporal parameters through a multi-view video camera, selecting the electrical stimulation strategy with the greatest difference from natural gait parameters at different viewing angles, and integrating it into the system control logic; S6) During walking, by regulating the function of specific lower limb muscles with electrical stimulation, the joint movement angle is affected, and the spatiotemporal and kinematic parameters of gait are changed, thereby achieving a gait feature with a camouflage effect.

2. A gait camouflage method for counteracting biological recognition according to claim 1, characterized in that: In step S1), the gait camouflage device includes an inertial sensor module, an electrical stimulation electrode module, a main control unit, and a power module. The main control unit is connected to the inertial sensor module, the electrical stimulation electrode module, and the power module, respectively. The power module is used to supply power to the inertial sensor module, the electrical stimulation electrode module, and the main control unit.

3. The gait camouflage method for resisting biological recognition according to claim 2, characterized in that: In step S1), the inertial sensor module is worn on the key parts of the user's lower limbs; the electrical stimulation electrode module is attached to the user's tibialis anterior and gastrocnemius muscles; the inertial sensor module is used to collect three-axis acceleration characteristics, three-axis angular velocity characteristics and magnetic field signal data; and the collected data is sent to the main control unit in real time through the SPI protocol; the main control unit determines the gait stage through the embedded algorithm, and controls the electrical stimulation electrode module to generate electrical stimulation pulses with corresponding parameters.

4. The gait camouflage method for resisting biological recognition according to claim 3, characterized in that: In step S1), the main control unit adopts a single-chip microcomputer with an ARM architecture; the electrical stimulation electrode module utilizes a rectangular wave generating circuit and a boost circuit to generate rectangular wave electrical pulses with different pulse widths and frequencies, which are output to the tibialis anterior and gastrocnemius muscle belly through hydrogel electrodes for functional electrical stimulation.

5. The gait camouflage method for resisting biological recognition according to claim 3, characterized in that: The inertial sensor module includes a three-axis acceleration sensor, a three-axis angular velocity sensor and a three-axis magnetometer.

6. The gait camouflage method for resisting biological recognition according to claim 2, characterized in that: In step S1), the power module includes a 3.7V rechargeable lithium battery, a boost circuit, a buck circuit, and multiple connection interfaces; the boost circuit is used to increase the 3.7V voltage to 5V and power the main control unit and the electrical stimulation electrode module; the buck circuit steps down the 3.7V voltage to 3.3V to power the inertial sensor module.

7. The gait camouflage method for resisting biological recognition according to claim 1, characterized in that: In step S3), the sample super vector of the current time segment is obtained by fusion, which specifically includes the following steps: S31), assign a unique name supervector D to each N feature of the kinematic signal i , and discretize the real value at the current moment and map it into value supervectors V of different levels i ; S32), name the super vector D i With value supervector V i A single feature supervector is generated by XOR operation, and all single feature supervectors are superimposed to form a synthetic supervector of the signal type; S33), performing temporal coding on the unimodal supervector through a bit shift operation ρ; S33), the synthetic supervectors of different signal types are fused through the concatenation operation Concat to obtain the sample supervector F corresponding to the current time segment; the fusion expression is: Where F is the sample supervector; T is the number of time slices; ρ is the bit shift operation; Concat is the concatenation operation; represents the i-th acceleration; represents the j-th angular velocity feature; represents the kth magnetic field signal type; n1, n2, n3 represent the number of acceleration features, angular velocity features, and magnetic field signal type features, respectively; represents the value supervector of the i-th acceleration feature at the t-th time segment, The supervector representing the value of the j-th angular velocity feature at the t-th time segment, A supervector representing the value of the k-th magnetic field signal feature at the t-th time segment; Represents the exclusive OR operation.

8. The gait camouflage method for resisting biological recognition according to claim 1, characterized in that: In step S4), the sample supervector is input into the hyperdimensional computing framework for real-time classification of gait phases, which specifically includes the following steps: S41), according to the gait stage label, the numerical superposition of each element of the motion sample supervector G of different time segments under the same label is performed to obtain a new supervector with unchanged dimension; S42), performing a sign function binarization on each element of the new supervector to generate a gait phase reference supervector M whose elements are all 1 or -1; S43), calculate the Hamming distance between the sample super vector G and the gait stage reference vector M through an XOR operation, and use it as a classification basis to predict the gait stage at the current moment, that is: Where, d H (G,M) represents the Hamming distance between G and M, f i and m i are the elements of G and M at the ith position, symbol Represents the exclusive-or operation.

9. The gait camouflage method for resisting biological recognition according to claim 1, characterized in that: In step S5), the functional electrical stimulation strategy with the largest difference from the natural gait parameters is collected by a multi-view video camera and selected to be incorporated into the control logic, which specifically includes the following steps: S51), the user wears the gait camouflage device and walks, applying electrical stimulation of different positions, intensities and durations to the lower limb muscles, and establishing a projection relationship between the electrical stimulation parameters and various gait characteristics; S52), collecting lower limb kinematics and spatiotemporal parameters through multi-view video cameras, and quantitatively evaluating the effect of electrical stimulation on natural gait patterns; S53), taking natural gait as a reference, extracting lower limb movement characteristics, constructing gait parameter characteristics corresponding to different types of gaits, adjusting the electrical stimulation strategy through a correspondence table between electrical stimulation parameters and gait characteristics, and simulating gait characteristics of different ages, genders or body shapes; S54), comparing the motion models under different electrical stimulation strategies with the natural gait, selecting the strategy with the greatest difference from the natural gait, and integrating it into the system control as the optimal electrical stimulation scheme.

10. The gait camouflage method for resisting biological recognition according to claim 1, characterized in that: In step S6), the gait motion parameters are adjusted during walking to achieve camouflage, which specifically includes the following steps: S61), applying electrical stimulation pulses to the tibialis anterior muscle and the gastrocnemius muscle through the electrical stimulation electrode module; S62) During normal walking, electrical stimulation is applied to the gastrocnemius muscle during the swing phase and to the tibialis anterior muscle during the support phase to adjust the joint angle and change the gait parameters to achieve a camouflage effect.

Images