Method, device and system for identity recognition by extracting bone features through knuckle knocking sound

Abstract

The present invention is a method, device and system for extracting bone features through the sound of knuckle knocking to perform identity recognition. The method includes the following steps: 1) sound collection: collect the sound signal generated by the user's knuckle knocking through a high-sensitivity microphone; 2) preprocessing: perform preprocessing operations on the collected sound; 3) feature extraction; 4) model training: perform model training on the extracted feature vector, using a supervised learning algorithm; 5) classification and recognition: establish the bone feature parameters and the user's identity, use the trained model to classify the extracted bone features, and output the user identity verification result; 6) database and model update: used to store the user's bone feature template and model parameters, supporting offline update and dynamic expansion; 7) practical application. The present invention realizes accurate, highly private and widely environmentally adaptable identity authentication through the characteristics that the sound signal generated when the knuckle knocking is highly correlated with the individual bone features.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence identity recognition technology, and specifically to a method, device and system for identity recognition by extracting bone features through the sound of knuckle knocking. Background Art

[0002] At present, identity recognition, as an important identification technology related to individual information security, plays an indispensable role in various industries. How to accurately establish the corresponding association information between an individual and his or her identity information is an important issue facing today's society. Nowadays, with the progress of some research on biometric extraction, biometrics such as facial features, iris patterns, and fingerprints have been widely used in identity recognition applications, but these are ultimately about the recognition of external human biometrics, and in actual applications are limited by application scenarios, and there is also a risk of user privacy exposure.

[0003] Current identity recognition technologies include fingerprint, face, iris, voiceprint and other methods. Although these methods are widely used, they have their own limitations: fingerprint recognition is easily affected by the surface state of the finger (such as humidity and contamination); face recognition is not effective in low light or complex backgrounds; iris recognition is costly and has a poor user experience; voiceprint recognition is easily affected by environmental noise and user voice changes. To this end, a method, device and system for extracting bone features from the sound of knuckle knocking for identity recognition are designed to overcome the above problems. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art and to provide a method, device and system for identity recognition by extracting bone features through the sound of knuckle knocking, so as to achieve accurate identity authentication with high privacy security and wide environmental adaptability.

[0005] The present invention is implemented by the following technical solution: a method for identity recognition by extracting bone features through the sound of knuckle knocking, the method comprising the following steps: 1) Sound collection: The sound signal generated by the user's knuckle tapping is collected through a high-sensitivity microphone; 2) Preprocessing: preprocess the collected sound; 3) Feature extraction: Use signal processing technology and machine learning methods to extract skeleton features from the knocking sound; 4) Model training: Perform model training on the extracted feature vectors using a supervised learning algorithm; 5) Classification and recognition: Establish the skeleton feature parameters and user identity, use the trained model to classify the extracted skeleton features, and output the user identity verification results; 6) Database and model update: used to store the user's bone feature template and model parameters, supporting offline update and dynamic expansion; 7) Practical application.

[0006] As a preferred embodiment, the pre-processing method in step 2) is to perform noise reduction on the collected knocking sound, and the specific noise reduction steps are: 1) Environmental noise filtering: Use a bandpass filter to remove environmental noise; 2) Door material adaptation: Establish acoustic characteristic models of various door materials to eliminate interference caused by door materials in knocking sounds; Knock force normalization: Adjust the impact of knock force based on signal amplitude.

[0007] As a preferred embodiment, the specific method of extracting bone features in step 3) is: 1) Spectral analysis: extract the main frequency and harmonic characteristics of the knocking sound; 2) Time domain features: extract the amplitude change, duration and other features of the knock signal; 3) Individual difference characteristics: Finally, the tapping force is normalized to obtain characteristic parameters related to the bone structure.

[0008] Preferably, the specific algorithm model in step 4) includes a convolutional neural network or a support vector machine to optimize the performance of the classifier.

[0009] A device for identity recognition by extracting bone features from the sound of knocking on a door with knuckles, comprising: Sound collection module: used to collect the sound signal generated by the user's knuckle tapping; Preprocessing module: performs noise reduction on the collected knocking sound; Feature extraction module: used to extract the corresponding features of the sound after noise reduction processing; Model training module: Perform model training on the extracted feature vectors using supervised learning algorithms; Classification and recognition module: Use the trained model to classify the extracted bone features; Database and model update module: used to store the user's bone feature template and model parameters, supporting offline updates and dynamic expansion.

[0010] A smart door lock identity authentication system is installed in a smart door lock. The identity authentication system adopts the above-mentioned device or method to complete identity authentication and unlock the door by the user knocking on the door. If there is a visitor knocking on the door who has been input in advance, the user can be informed of the visitor in advance, thereby improving the safety and intelligence level of residence.

[0011] An identity authentication system in a financial system, the system is used in a financial authentication system, the financial authentication system adopts the above-mentioned device or method to extract the customer's Google feature security identity authentication through knocking sound, thereby improving security and user experience.

[0012] The beneficial effects of the present invention are as follows: Compared with the prior art, the present invention has improvements in many aspects: the unique performance of bone features is used to ensure reliable and high-precision identity authentication. The present invention uses multiple audio noise reduction and modeling of door materials to effectively reduce the influence of factors such as environmental noise, door material differences, and knocking force. Since bone features are difficult to forge and do not need to rely on external biometric images, user privacy can be effectively protected. It has a wider range of application scenarios and can be applied to access control systems, smart homes, financial authentication and other scenarios. The user experience is highly comfortable, and identity authentication is achieved directly by knocking on the door without the need for additional equipment or complex operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 This is a flow chart of the preprocessing module, feature extraction module and deep learning framework in the present invention. DETAILED DESCRIPTION

[0014] In order to enable those skilled in the art to more clearly understand the objectives, technical solutions and advantages of the present invention, the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0015] In the description of the present invention, it is necessary to understand that the orientation or position relationship indicated by the terms such as "up", "down", "left", "right", "inside", "outside", "horizontal" and "vertical" is based on the orientation or position relationship shown in the accompanying drawings, and is only for the convenience of describing the present invention, rather than indicating or implying that the device or original referred to must have a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0016] The present invention will be described in detail below with reference to the accompanying drawings: Figure 1 As shown, a method for identity recognition by extracting bone features through the sound of knuckle knocking, the method comprises the following steps: 1) Sound collection: The sound signal generated by the user's knuckle tapping is collected through a high-sensitivity microphone; 2) Preprocessing: preprocess the collected sound; 3) Feature extraction: Use signal processing technology and machine learning methods to extract skeleton features from the knocking sound; 4) Model training: Perform model training on the extracted feature vectors using a supervised learning algorithm; 5) Classification and recognition: Establish the skeleton feature parameters and user identity, use the trained model to classify the extracted skeleton features, and output the user identity verification results; 6) Database and model update: used to store the user's skeleton feature template and model parameters, supporting offline update and dynamic expansion; 7) Practical application.

[0017] As a preferred embodiment, the pre-processing method in step 2) is to perform noise reduction on the collected knocking sound, and the specific noise reduction steps are: 1) Environmental noise filtering: Use a bandpass filter to remove environmental noise; 2) Door material adaptation: Establish acoustic characteristic models of various door materials to eliminate interference caused by door materials in knocking sounds; Knock force normalization: Adjust the impact of knock force based on signal amplitude.

[0018] As a preferred embodiment, the specific method of extracting bone features in step 3) is: 1) Spectral analysis: extract the main frequency and harmonic characteristics of the knocking sound; 2) Time domain features: extract the amplitude change, duration and other features of the knock signal; 3) Individual difference characteristics: Finally, the tapping force is normalized to obtain characteristic parameters related to the bone structure.

[0019] Preferably, the specific algorithm model in step 4) includes a convolutional neural network or a support vector machine to optimize the performance of the classifier.

[0020] A device for identity recognition by extracting bone features from the sound of knocking on a door with knuckles, comprising: Sound collection module: used to collect the sound signal generated by the user's knuckle tapping; Preprocessing module: performs noise reduction on the collected knocking sound; Feature extraction module: used to extract the corresponding features of the sound after noise reduction processing; Model training module: Perform model training on the extracted feature vectors using supervised learning algorithms, such as convolutional neural networks (CNN) or support vector machines (SVM), to optimize the performance of the classifier.

[0021] Classification and recognition module: Use the trained model to classify the extracted bone features; Database and model update module: used to store the user's bone feature template and model parameters, supporting offline updates and dynamic expansion.

[0022] A smart door lock identity authentication system is installed in a smart door lock. The identity authentication system adopts the above-mentioned device or method to complete identity authentication and unlock the door by the user knocking on the door. If there is a visitor knocking on the door who has been input in advance, the user can be informed of the visitor in advance, thereby improving the safety and intelligence level of residence.

[0023] An identity authentication system in a financial system, the system is used in a financial authentication system, the financial authentication system adopts the above-mentioned device or method to extract the customer's Google feature security identity authentication through knocking sound, thereby improving security and user experience.

[0024] There is a small distance between the bones of finger joints and the surface of the skin. The sound when knocking on the door will be affected by the state of the bones under the skin of the finger joints (bone density, length, thickness, shape), and these features are unique. The method, device and system designed by the present invention for identity recognition by extracting bone features from the sound of knocking on the door with the knuckles focuses on extracting bone features from the knocking sound to uniquely identify the user's identity, thereby achieving accurate and secure identity authentication, while simplifying the identity recognition process to increase speed and the scope of applicable scenarios.

[0025] Anyone familiar with the technology may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by a person of ordinary skill in the art without departing from the spirit and technical concept disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. A method for identity recognition by extracting bone features from the sound of knuckle knocking, characterized in that: The method comprises the following steps: 1) Sound collection: The sound signal generated by the user's knuckle tapping is collected through a high-sensitivity microphone; 2) Preprocessing: preprocess the collected sound; 3) Feature extraction: Use signal processing technology and machine learning methods to extract skeleton features from the knocking sound; 4) Model training: Perform model training on the extracted feature vectors using a supervised learning algorithm; 5) Classification and recognition: Establish the skeleton feature parameters and user identity, use the trained model to classify the extracted skeleton features, and output the user identity verification results; 6) Database and model update: used to store the user's skeleton feature template and model parameters, supporting offline update and dynamic expansion; 7) Practical application.

2. The method according to claim 1, characterized in that: The pre-processing method in step 2) is to perform noise reduction on the collected knocking sound, and the specific noise reduction steps are: 1) Environmental noise filtering: Use a bandpass filter to remove environmental noise; 2) Door material adaptation: Establish acoustic characteristic models of various door materials to eliminate interference caused by door materials in knocking sounds; Knock force normalization: Adjust the impact of knock force based on signal amplitude.

3. The method according to claim 1, characterized in that: The specific method of extracting bone features in step 3) is: 1) Spectral analysis: extract the main frequency and harmonic characteristics of the knocking sound; 2) Time domain features: extract the amplitude change and duration features of the knock signal; 3) Individual difference characteristics: Finally, the tapping force is normalized to obtain characteristic parameters related to the bone structure.

4. The method according to claim 1, characterized in that: The specific algorithm model in step 4) includes a convolutional neural network or a support vector machine to optimize the performance of the classifier.

5. A device for identity recognition by extracting bone features from the sound of knocking on a door with knuckles, characterized in that: include: Sound collection module: used to collect the sound signal generated by the user's knuckle tapping; Preprocessing module: performs noise reduction on the collected knocking sound; Feature extraction module: used to extract the corresponding features of the sound after noise reduction processing; Model training module: Perform model training on the extracted feature vectors using supervised learning algorithms; Classification and recognition module: Use the trained model to classify the extracted bone features; Database and model update module: used to store the user's bone feature template and model parameters, supporting offline updates and dynamic expansion.

6. A smart door lock identity authentication system, which is installed in a smart door lock, characterized in that: The identity authentication system adopts the device or method described in claim 1 or 5, and completes identity authentication and unlocking by the user knocking on the door. If there is a visitor knocking on the door with a pre-entered input, the user can be informed of the visitor in advance, thereby improving the level of residential safety and intelligence.

7. An identity authentication system in a financial system, the system being used in a financial authentication system, characterized in that: The financial authentication system adopts the device or method described in claim 1 or 5 to extract the customer's Google feature security identity authentication through knocking sound, thereby improving security and user experience.

Images