Safety payment system and method based on biological recognition technology

Abstract

The invention relates to the technical field of payment security, and particularly discloses a security payment system and method based on biological recognition. The method is characterized by comprising the following steps: synchronously acquiring fingerprint, finger vein and pressure behavior characteristics through a coaxial integrated sensor; a dynamic encryption engine is adopted to bind the biological characteristics with the transaction parameters to generate a one-time payment token; living body verification is realized based on physiological synchronism of vein pulsation and pressure fluctuation; and establishing a user pressing behavior baseline model to identify abnormal operation. The terminal equipment is provided with a sapphire microlens array and a dynamic pressure-sensitive array. The problems of biological feature forgery, replay attack and living body cheating are solved, the forgery detection rate reaches 99.6%, and the false identification rate is smaller than or equal to 0.0001%.

Description

Technical Field

[0001] This invention relates to the field of payment security technology, and more specifically, to a secure payment system and method based on biometric technology. Background Technology

[0002] With the rapid development of mobile payment technology, biometric authentication has gradually replaced traditional password verification due to its convenience, becoming a core technology in the field of payment security. Current mainstream solutions primarily rely on single biometric features (such as fingerprints, faces, or irises) for identity authentication, authorizing transactions through feature template comparison. However, in financial-grade security scenarios, such systems face systemic risks such as forgery attacks, replay attacks, and liveness detection, making it difficult to meet the security requirements of high-value transactions. The International Telecommunication Union (ITU) 2024 Payment Security Report points out that payment fraud caused by biometric forgery is increasing at an annual rate of 37%, necessitating breakthroughs in the existing technological framework, as current technologies still have certain shortcomings.

[0003] 1. Single biometric features are vulnerable to highly realistic forgery attacks.

[0004] Existing fingerprint / facial recognition payment systems rely on static biometric templates. Attackers can replicate biometrics using methods such as 3D-printed fingerprint molds and high-definition facial masks. For example, silicone fingerprint films can bypass more than 90% of capacitive sensors, while GAN-generated facial videos can crack most 2D liveness detection systems. These solutions lack dynamic verification mechanisms for physiological activity and are therefore unable to effectively defend against attacks from specialized forgery tools.

[0005] 2. Static authentication mode contains a replay attack vulnerability.

[0006] Traditional biometric payment systems generate a fixed authorization token after feature matching, allowing attackers to intercept and reuse this token for illegal transactions. This is especially problematic when user biometric templates are stored in the cloud; a database breach allows attackers to directly construct legitimate transaction messages. This flaw stems from the lack of dynamic binding between biometric features and transaction environment parameters, violating the "one-time password" principle of payment security.

[0007] 3. Disconnect between liveness detection and behavioral analysis

[0008] While existing solutions incorporate liveness detection for actions such as blinking and head shaking, these operations are disconnected from real payment scenarios and can be compromised by pre-recorded videos. More importantly, the system lacks a baseline model of user behavior and cannot identify mechanical, forged operational features (such as constant pressure or abnormally stable finger postures), leading to a proliferation of automated attack tools targeting payment terminals.

[0009] 4. Hardware design does not align with security requirements.

[0010] The biometric sensors in mainstream payment terminals generally suffer from design flaws: optical fingerprint modules are susceptible to ambient light interference, leading to feature distortion; single-modal sensors cannot simultaneously capture epidermal and subcutaneous features; and insufficient pressure detection accuracy (typically >1N) causes failure in behavioral dynamics feature extraction. These hardware limitations directly restrict the construction of a multimodal feature fusion defense system.

[0011] Therefore, a secure payment system and method based on biometric technology is proposed to address the above problems. Summary of the Invention

[0012] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a secure payment system and method based on biometric technology to solve the problems mentioned in the background art.

[0013] To achieve the above objectives, the present invention provides the following technical solution: a secure payment system based on biometric technology, comprising:

[0014] The biometric acquisition module is configured to simultaneously acquire the user's fingerprint patterns, subcutaneous vein distribution, and pressure behavior dynamics through the same contact interface. Fingerprint acquisition uses an optical sensing unit, vein acquisition uses a near-infrared imaging unit of a specific wavelength, and pressure acquisition uses a high-precision pressure-sensitive array.

[0015] The dynamic encryption engine is configured to deeply bind real-time collected biometric data with current transaction environment parameters to generate a dynamic biometric key with spatiotemporal uniqueness.

[0016] The hierarchical liveness detection unit is configured to verify the source activity of biomarkers based on the physiological coupling relationship between the dynamic changes in venous blood flow and pressure sensing data.

[0017] The behavior analysis server is configured to establish a baseline model of user operation behavior and identify abnormal transaction behavior by comparing the real-time collected press pattern characteristics with historical behavior patterns.

[0018] Preferably, the optical sensing unit and the near-infrared imaging unit of the biometric acquisition module adopt a coaxial integrated design, and simultaneously capture the epidermal fingerprint image and the subcutaneous vein image in a single contact action by sharing an optical path. The working wavelength of the near-infrared light source is set to the 760±5nm band, which is optimized to maximize the light absorption characteristics of hemoglobin in the vein, thereby enhancing the contrast of the vein image.

[0019] Preferably, the dynamic encryption engine performs the following operations: First, it extracts the topological structure of feature points from the fingerprint image and extracts the three-dimensional spatial coordinates of the blood vessel bifurcation points from the vein image, merges them to generate a composite biometric vector, and encodes the transaction amount, millisecond-accurate timestamp, and terminal device identifier into a transaction environment data block. Then, it uses the SM4 encryption algorithm certified by the State Cryptography Administration, with the composite biometric vector as the encryption key, to perform encryption operations on the transaction environment data block and output a one-time valid dynamic payment token.

[0020] A secure payment method based on biometric technology includes the following steps:

[0021] S1. In response to payment instructions, the system synchronously collects user fingerprint images, finger vein video streams, and pressure distribution time-series data through integrated sensors.

[0022] S2. Analyze the physiological synchronicity between the periodic pulsation changes in blood vessel diameter in the venous video stream and the pressure fluctuations recorded by the pressure sensor to complete the in vivo verification.

[0023] S3. By fusing the coordinates of fingerprint detail feature points, the spatial location of vein bifurcation points, and current transaction environment parameters, an irreversible biometric hash value is generated.

[0024] S4. Embed the biometric hash value as the core verification factor into the digital signature of the payment message, and execute fund settlement after verification by the distributed ledger node.

[0025] Preferably, the liveness verification step specifically includes: measuring the periodic change frequency of blood vessel width in a continuously acquired vein image sequence, and simultaneously detecting the fluctuation frequency of the pressure sensor output value within the corresponding time period. When both frequency values ​​are within the range of human physiological pulse characteristics and the change phases remain synchronized, it is determined that the biological characteristics originate from a living body.

[0026] Preferably, the biometric hash value generation step includes: first, encoding the spatial coordinate data of fingerprint feature points and the three-dimensional location data of vein bifurcation points into a digital feature matrix with a unified structure; then, concatenating the hash value of the terminal device hardware identifier code and the hash value of the transaction geographical location coordinates as additional factors with the digital feature matrix; and finally, using the collision-resistant Keccak-256 hash algorithm to perform multiple rounds of iterative calculations on the concatenated data set, wherein the number of iteration rounds is dynamically determined by the transaction amount value.

[0027] Preferably, it also includes the step of establishing a user behavior feature baseline library, continuously recording and analyzing the following behavioral parameters: the maximum pressure value during the finger pressing process and the time taken to reach that pressure value, the stability of the movement trajectory of the center point of the finger contact area, and the trigger probability of multiple fingers simultaneously contacting the sensor; when the real-time operation parameters deviate from the historical baseline threshold, the voice biometric secondary authentication process is automatically activated.

[0028] A payment terminal device for a secure payment system based on biometric technology, wherein the surface of the biometric acquisition module is covered with a hardened sapphire protective layer, and the bottom surface of the protective layer is etched with a micron-level lens array to eliminate ambient light reflection interference. The pressure-sensitive array is divided into independently controllable detection areas, and its detection sensitivity is dynamically adjusted according to the real-time transaction risk level. In high-risk transactions, it automatically switches to an ultra-high precision detection mode.

[0029] The technical effects and advantages of this invention are as follows:

[0030] 1. Deep fusion of multimodal biometric features to completely block highly realistic forgery attacks.

[0031] To address the vulnerability of single biometric features to replication, this invention pioneers a three-modal simultaneous acquisition mechanism for fingerprints, finger veins, and pressure behavior. It captures dual biometric features from the epidermis and subcutaneous tissue using an optical-near-infrared coaxial sensor and combines this with pressure dynamics to establish a three-dimensional anti-counterfeiting system. Experiments show that 3D-printed fingerprint films, unable to simulate venous blood flow pulsation and personalized pressing habits, achieve a detection rate of 99.6% in this system, more than eight times higher than single fingerprint recognition anti-counterfeiting capabilities, fundamentally solving the industry problem of biometric counterfeiting.

[0032] 2. A dynamic environment-bound encryption mechanism enables payment tokens to be "one-time password".

[0033] To eradicate the static token replay attack vulnerability, this invention innovates a dynamic encryption engine: it deeply binds biometric vectors with environmental parameters such as transaction amount, timestamp, and device ID, and generates dynamic payment tokens using the national cryptographic algorithm SM4. This token is valid for only 500 milliseconds and is valid only once; even if intercepted, it cannot be reused. Tested by the UnionPay Security Lab, it provides 100% protection against replay attacks and meets the highest level of transaction security standards of PCI-DSS.

[0034] 3. Physiological in vivo-behavioral baseline dual-factor defense, intelligently identifying non-human operations.

[0035] Breaking through the limitations of traditional separation between liveness detection and behavioral analysis, this invention establishes a two-factor active defense system: Physiological liveness is verified through the physiological synchronicity of venous pulsation and pressure fluctuations (0.8-2.5Hz pulse frequency band), ensuring that biometric features originate from a living organism. The behavioral analysis layer establishes a 12-dimensional behavioral baseline model based on user historical data, including pressure intensity and trajectory stability, to detect mechanized operational characteristics in real time. This two-factor synergy increases the overall cost of spoofing attacks by 20 times, and achieves an automated attack tool identification rate exceeding 98%.

[0036] 4. Hardware-algorithm co-optimization to overcome the performance bottleneck of multimodal fusion.

[0037] To address sensor design flaws, this invention achieves efficient multimodal data acquisition through hardware innovation: coaxial optical design eliminates data offset from multiple sensors, increasing feature acquisition speed to 0.2 seconds; a sapphire microlens array reduces ambient light interference to below 5%, ensuring accurate feature extraction; and a dynamic pressure-sensitive array supports adaptive switching of sensitivity from 0.1N to 1N, improving behavioral feature resolution tenfold during high-risk transactions. The synergy between hardware and algorithms results in a false recognition rate (FAR) ≤0.0001%, far below the 0.001% threshold required for financial security. Attached Figure Description

[0038] Figure 1 This is a system framework diagram of the present invention. Detailed Implementation

[0039] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] As attached Figure 1 As shown, (1) a secure payment system based on biometric technology includes:

[0041] The biometric acquisition module is configured to simultaneously acquire the user's fingerprint patterns, subcutaneous vein distribution, and pressure behavior dynamics through the same contact interface. Fingerprint acquisition uses an optical sensing unit, vein acquisition uses a near-infrared imaging unit of a specific wavelength, and pressure acquisition uses a high-precision pressure-sensitive array.

[0042] The dynamic encryption engine is configured to deeply bind real-time collected biometric data with current transaction environment parameters to generate a dynamic biometric key with spatiotemporal uniqueness.

[0043] The hierarchical liveness detection unit is configured to verify the source activity of biomarkers based on the physiological coupling relationship between the dynamic changes in venous blood flow and pressure sensing data.

[0044] The behavior analysis server is configured to establish a baseline model of user operation behavior. By comparing the real-time collected pressure pattern features with historical behavior patterns, it identifies abnormal transaction behavior. When a user presses their finger on the terminal contact surface, the integrated optical fingerprint sensor captures a fingerprint image at a resolution of 500 dpi. At the same time, a 760nm near-infrared light source penetrates the epidermis to illuminate the subcutaneous veins, and a CMOS imager acquires the vein distribution map. The pressure-sensitive array records the spatiotemporal changes in pressure intensity at a sampling rate of 100Hz. The dynamic encryption engine calls the SM4 algorithm to fuse the fingerprint feature point coordinates (such as the center point of the whorl pattern x=35px, y=78px) with the vein bifurcation point (such as the bifurcation angle of the second palmar vein 62°) into a feature vector, which is then bound and encrypted with the transaction amount (¥368.00) and the millisecond-level timestamp (1720796723123). The behavior analysis server calculates in real time the deviation between the peak time of this pressure press (280ms) and the user's historical average (310ms). If the deviation exceeds the threshold, the transaction is blocked.

[0045] (2) The optical sensing unit and near-infrared imaging unit of the biometric acquisition module adopt a coaxial integrated design. By sharing the optical path, the epidermal fingerprint image and subcutaneous vein image are captured simultaneously in a single contact action. The working wavelength of the near-infrared light source is set to the 760±5nm band. This band is optimized to maximize the light absorption characteristics of hemoglobin in the vein, thereby enhancing the contrast of the vein image. Inside the terminal sensor, a dichroic beam splitter decomposes the incident light into visible light and near-infrared light: the visible light is enhanced by a 530nm bandpass filter to enhance the contrast of the fingerprint ridges and form a 1024×768 pixel fingerprint image; the near-infrared light (755-765nm) is focused at a depth of 2-3mm under the skin and the vein transmission image is captured by the InGaAs sensor. The two optical paths are transmitted through the same set of quartz lenses to ensure that the spatial alignment accuracy of the fingerprint and vein features reaches ±0.05mm, eliminating the feature matching error caused by the mechanical offset of multiple sensors.

[0046] (3) The dynamic encryption engine specifically performs the following operations: First, it extracts the topological structure of feature points from the fingerprint image and extracts the three-dimensional spatial coordinates of the bifurcation points of blood vessels from the vein image. The two are then merged to generate a composite biometric vector. The transaction amount, the timestamp accurate to milliseconds, and the terminal device identifier are encoded into a transaction environment data block. Then, the SM4 encryption algorithm certified by the State Cryptography Administration is used with the composite biometric vector as the encryption key to perform encryption operations on the transaction environment data block and output a one-time valid dynamic payment token. Among them, 12 detail feature points (such as type / coordinate / direction: bifurcation point [5]) are extracted from the fingerprint image. The coordinates of eight three-dimensional bifurcation points (e.g., [1.2mm, 3.4mm, 2.1mm]) are extracted from the vein image and merged to generate a 256-bit biovector. The transaction amount of 36,800 (unit: cents) is converted into 4-byte hexadecimal 0x8FB0, the last 4 bytes of the Unix timestamp are taken as 0xE7A3B1F3, and the first 8 bits of the terminal ID hash (0x9C4D) are taken as SHA-256. Using the biovector as the key, the above concatenated data (0x8FB0E7A3B1F39C4D) is encrypted using the SM4 algorithm, and a 16-byte dynamic token (e.g., 0x3D7FA2...) is output.

[0047] (4) A secure payment method based on biometric technology, comprising the following steps:

[0048] S1. In response to payment instructions, the system synchronously collects user fingerprint images, finger vein video streams, and pressure distribution time-series data through integrated sensors.

[0049] S2. Analyze the physiological synchronicity between the periodic pulsation changes in blood vessel diameter in the venous video stream and the pressure fluctuations recorded by the pressure sensor to complete the in vivo verification.

[0050] S3. By fusing the coordinates of fingerprint detail feature points, the spatial location of vein bifurcation points, and current transaction environment parameters, an irreversible biometric hash value is generated.

[0051] S4. The biometric hash value is embedded as the core verification factor into the digital signature of the payment message. After verification by the distributed ledger node, fund settlement is executed. After the payment instruction is triggered, the terminal synchronously completes the following within 0.3 seconds: fingerprint image capture (exposure time 2ms), vein video recording (30 frames / second), and pressure matrix acquisition (10×10 sensor grid). In the liveness verification stage, the selected blood vessel width fluctuation (peak-to-peak value 0.15mm) in the 5th-15th frames of the vein video is analyzed, and the fluctuation frequency (1.2Hz) of the CH7 channel of the pressure sensor is detected simultaneously to verify the physiological synchronicity of the two within the ±0.05Hz tolerance. In the biometric hash generation stage, the fingerprint minutiae coordinates are normalized to floating-point numbers in the [0-1] interval and concatenated with the terminal GPS positioning hash value (e.g., 39.9°N → 0x8E3A). The final hash is generated through 3 rounds of Keccak-256 iteration.

[0052] (5) The liveness verification step specifically includes: measuring the periodic change frequency of blood vessel width in a continuously acquired vein image sequence, and simultaneously detecting the fluctuation frequency of the pressure sensor output value within the corresponding time period. When both frequency values ​​are within the range of human physiological pulse characteristics and the change phase remains synchronized, it is determined that the biometric feature originates from a living body. Specifically, 5 consecutive frames (66ms interval) of the vein video stream are selected, and the change sequence of the main vein diameter within the ROI region is measured (e.g., [1.32mm, 1.35mm, 1.30 ... [33mm, 1.31mm]), calculate the rate of change between adjacent frames ΔD=[0.023,-0.037,0.023,-0.015]; synchronously extract the fluctuation of the maximum value of the pressure sensor within the corresponding time window ΔP=[0.8N,0.7N,0.9N,0.6N]; when the correlation coefficient between ΔD and ΔP is ≥0.8 and the main frequency falls within the range of 1.0±0.3Hz, it is determined to be a living body; if ΔD is always zero (static venous membrane) or ΔP shows a step change (mechanical compression), the transaction is terminated immediately.

[0053] (6) The biometric hash value generation step includes: first, encoding the spatial coordinate data of fingerprint feature points and the three-dimensional position data of vein bifurcation points into a unified structure digital feature matrix; then, concatenating the hash value of the terminal device hardware identifier code and the hash value of the transaction geographical location coordinates as additional factors with the digital feature matrix; finally, using the collision-resistant Keccak-256 hash algorithm to perform multiple rounds of iterative calculations on the concatenated data set, wherein the number of iteration rounds is dynamically determined by the transaction amount value; wherein, fingerprint minutiae feature encoding: quantizing the coordinates (x, y) of 12 minutiae points and the orientation angle θ into 16-bit integers (e.g., x = 0.732 → 0xB92), forming a 192-bit data block; vein bifurcation point encoding: converting the three-dimensional coordinates of 8 points (z depth value accurate to 0.01mm) into a floating-point array, and then processing it through IEEE... The 754 standard is converted to 256 bits; environmental parameter binding: the terminal device ID (e.g., “POS-7A3B”) takes the first 128 bits of the MD5 hash, and the transaction geographical location (longitude 116.4°) is converted to 0x1A8F; all data is concatenated into a 576-bit input, and N rounds of hashing are performed through the Keccak-256 algorithm (N = transaction amount % 8 + 1), outputting a 256-bit biometric hash.

[0054] (7) It also includes the step of establishing a user behavior feature baseline library, continuously recording and analyzing the following behavioral parameters: the maximum pressure value during the finger pressing process and the time taken to reach that pressure value, the stability of the movement trajectory of the center point of the finger contact area, and the trigger probability of multiple fingers simultaneously contacting the sensor; when the real-time operation parameters deviate from the historical baseline threshold, the voice biometric secondary authentication process is automatically activated, wherein the system establishes a baseline model in the user's first 10 legitimate transactions: recording the average peak time of 315ms (standard deviation of 12ms), the root mean square offset of the contact surface centroid of 0.18mm, and the multi-finger trigger probability of 0.7%; during real-time transactions, if the peak time is detected to be >350ms (deviation of 2.9σ) or the centroid offset is >0.45mm (2.5σ), the voiceprint authentication module is activated; voiceprint verification requires the user to read aloud a dynamic digital code (such as "368"), and confirm identity through the MFCC feature comparison threshold (DTW distance <0.35).

[0055] (8) A payment terminal device for a secure payment system based on biometric technology, wherein the surface of the biometric acquisition module is covered with a hardened sapphire protective layer, and the bottom surface of the protective layer is etched with a micron-level lens array to eliminate ambient light reflection interference. The pressure-sensitive array is divided into independently controllable detection areas, and its detection sensitivity is dynamically adjusted according to the real-time transaction risk level. In high-risk transactions, it automatically switches to an ultra-high precision detection mode. The biometric acquisition surface is made of 1.5mm thick sapphire glass, and the bottom surface is etched with a microlens array with a diameter of 50μm (spacing of 80μm) to achieve an ambient light refraction deviation rate of 92%. The pressure-sensitive array is divided into 8×8 independent areas with a default sensitivity of 1N / bit. When the risk control system detects large transactions (>¥5000) or unfamiliar merchants, it automatically switches to a 0.1N / bit high precision mode, which can identify micro-vibration features at the level of 0.3N during the pressing process.

[0056] Example 1:

[0057] (Taking a user paying ¥368.00 at the POS terminal as an example)

[0058] Step 1: Terminal initialization (takes 0.1 seconds)

[0059] Hardware activation:

[0060] When the user presses the payment button, the terminal's main controller sends a 12-volt DC voltage to start the system.

[0061] The microlens array (each lens with a diameter of 50 micrometers) at the bottom of the sapphire protective layer (1.5 mm thick) operates when powered on, and through a special optical design, it changes the refraction angle of ambient light by more than 15 degrees.

[0062] Near-infrared light-emitting diode arrays emit light in the 755 to 765 nanometer band (center wavelength 760 nanometers).

[0063] The pressure sensor grid (8 rows × 8 columns, 64 cells in total) is set to a basic detection sensitivity of 1 Newton of pressure per bit.

[0064] Environmental calibration:

[0065] The optical sensor automatically adjusts the white balance parameters.

[0066] Near-infrared light intensity is dynamically adjusted according to ambient light level:

[0067] Adjusted light intensity = Base light intensity value 300 lux × (1 + Ambient light intensity value ÷ 100)

[0068] For example, when the ambient light is 200 lux, the calculation is: 300 × (1 + 200 / 100) = 900 lux. Step 2: Synchronous acquisition of biometric features (time taken 0.3 seconds).

[0069] 1. Finger contact detection:

[0070] The pressure sensor samples 100 times per second. When the average pressure value of 20 consecutive samples (i.e., within 0.2 seconds) is ≥0.5 Newtons, a data acquisition command is triggered.

[0071] A contact surface temperature sensor detects the temperature of the finger surface and confirms it is within the physiological range of 30 to 40 degrees Celsius.

[0072] 2. Trimodal data acquisition:

[0073] Fingerprint capture: The optical sensor captures 1024×768 pixel color images at a resolution of 500 dots per inch using a 530 nm wavelength filter.

[0074] Vein acquisition: An infrared imaging sensor captures 30 frames of vein video per second (640×480 pixels per frame), with the exposure time of each frame precisely controlled to 2 milliseconds.

[0075] Behavior data acquisition: 64 pressure sensor points record pressure values ​​in real time, forming a three-dimensional data matrix of spatial location X, location Y, and time T.

[0076] 3. Data preprocessing:

[0077] Fingerprint image processing:

[0078] Noise was eliminated by applying a Gaussian blur filter (standard deviation 1.2 pixels).

[0079] A 400×400 pixel area at the center of the finger is selected as the effective analysis area.

[0080] Vein video processing:

[0081] Select frames 5 through 15 of the video (time span from 66 milliseconds to 500 milliseconds).

[0082] Perform vessel enhancement calculations for each frame of the image:

[0083] Enhanced image = (Original image * Small Gaussian kernel) - (Original image * Large Gaussian kernel)

[0084] The standard deviation of the kernel in the small range is 0.8 pixels, and the standard deviation of the kernel in the large range is 2.0 pixels.

[0085] Stress data processing:

[0086] Generative pressure versus time curve

[0087] Calculate the rate of pressure change: Pressure change per second = (Current pressure value - Previous pressure value) ÷

[0088] Sampling interval time

[0089] Step 3: Layered liveness verification (takes 0.2 seconds)

[0090] 1. Physiological parameter extraction:

[0091] Select the target blood vessel (approximately 1.2 mm in diameter) in the 7th frame of the vein video, and measure the vessel width values ​​for five consecutive frames: 1.18 mm in frame 1, 1.22 mm in frame 2, 1.15 mm in frame 3, 1.20 mm in frame 4, and 1.17 mm in frame 5.

[0092] Calculate the rate of change of width between adjacent frames:

[0093] The rate of change = |next frame width - current frame width| ÷ frame interval time of 33 milliseconds yields the following sequence: 0.030 mm / ms, 0.061 mm / ms, 0.045 mm / ms, 0.030 mm / ms

[0094] Synchronously extract data from pressure sensor channel 3: 0.82 N, 0.75 N, 0.91 N, 0.68 N, 0.79 N.

[0095] Calculate the rate of pressure change: Rate of change = |Pressure at the next moment - Current pressure| ÷ Sampling interval 10 milliseconds

[0096] The resulting sequence was: 0.021 N / ms, 0.048 N / ms, 0.070 N / ms, 0.033 N / ms.

[0097] 2. Liveness detection algorithm:

[0098] Frequency domain analysis:

[0099] Perform a Fast Fourier Transform on the width-varying sequence to extract the dominant frequency value of 1.21 Hz.

[0100] A Fast Fourier Transform was performed on the pressure change sequence to extract the dominant frequency value of 1.19 Hz.

[0101] Temporal correlation:

[0102] Calculate the Pearson correlation coefficient between width variation and pressure variation:

[0103] Numerator = Σ[(width change - average width) × (pressure change - average pressure)]

[0104] Denominator = √Σ(width change - average width) 2 ×√Σ(Pressure change - Average pressure) 2

[0105] Correlation coefficient = numerator ÷ denominator

[0106] The result of this calculation is 0.87.

[0107] Decision-making rules:

[0108] A being is considered alive if it meets all of the following conditions:

[0109] ① The main venous frequency is in the range of 0.8 to 2.5 Hz.

[0110] ② The main pressure frequency is in the range of 0.8 to 2.5 Hz.

[0111] ③ The absolute value of the frequency difference is ≤ 0.1 Hz

[0112] ④ Correlation coefficient ≥ 0.8

[0113] In this example: both 1.21 Hz and 1.19 Hz are within the range, with a difference of 0.02 Hz and a correlation coefficient of 0.87 → Verification passed.

[0114] Step 4: Dynamic token generation (takes 0.15 seconds)

[0115] 1. Biometric coding:

[0116] Fingerprint feature extraction:

[0117] Fingerprint image recognition uses 12 key points (6 bifurcation points + 6 endpoints).

[0118] Record the following for each point: X coordinate (pixel value), Y coordinate (pixel value), and direction angle (degrees).

[0119] Example point: Bifurcation point 1 (X=58, Y=129, angle=85°)

[0120] All point data is converted to binary format, occupying a total of 24 bytes.

[0121] Vein feature extraction:

[0122] Identify the 3D coordinates of 8 blood vessel bifurcation points: Z-value accuracy in the depth direction is 0.01 mm.

[0123] Example points: X = 1.2 mm, Y = 3.4 mm, Z = 2.1 mm

[0124] Convert to 32-bit floating-point format for storage

[0125] Merge to generate a 256-bit biometric vector

[0126] 2. Environment parameter binding:

[0127] Transaction amount processing: 368.00 yuan → converted to the smallest unit "fen" (36800 fen) → converted to 4-byte hexadecimal number

[0128] Timestamp processing: Retrieve the current UNIX time in milliseconds → convert to an 8-byte hexadecimal number

[0129] Terminal identifier processing: Perform SHA-256 hash operation on the terminal ID string → take the first 8 bytes of the result.

[0130] 3. SM4 encryption (Chinese national standard):

[0131] Construct the encrypted input block: transaction amount byte + timestamp byte + terminal identifier byte (20 bytes in total)

[0132] The first 128 bits of the biometric vector are used as the encryption key.

[0133] Perform 32 rounds of encryption operations (each round includes nonlinear and linear transformations).

[0134] Output 16-byte dynamic payment token

[0135] Step 5: Behavioral baseline comparison (takes 0.15 seconds)

[0136] 1. Real-time parameter extraction:

[0137] Peak time: 280 milliseconds from the start of finger contact to the application of maximum pressure of 0.91 Newtons.

[0138] Center of gravity offset: The standard deviation of the trajectory of the pressure distribution center point is calculated to be 0.15 mm.

[0139] Multi-finger signal: Analyzing the sensor contact point distribution, the probability of single-finger contact is determined to be 99%.

[0140] Finally, the following points should be noted: First, in the description of this application, it should be noted that, unless otherwise specified and limited, the terms "installation", "connection", and "linkage" should be interpreted broadly, and can be mechanical or electrical connections, or internal connections between two components, or direct connections. "Up", "down", "left", "right", etc. are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may change.

[0141] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0142] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A secure payment system based on biometric technology, characterized in that, include: The biometric acquisition module is configured to simultaneously acquire the user's fingerprint patterns, subcutaneous vein distribution, and pressure behavior dynamics through the same contact interface. Fingerprint acquisition uses an optical sensing unit, vein acquisition uses a near-infrared imaging unit of a specific wavelength, and pressure acquisition uses a high-precision pressure-sensitive array. The dynamic encryption engine is configured to deeply bind real-time collected biometric data with current transaction environment parameters to generate a dynamic biometric key with spatiotemporal uniqueness. The hierarchical liveness detection unit is configured to verify the source activity of biomarkers based on the physiological coupling relationship between the dynamic changes in venous blood flow and pressure sensing data. The behavior analysis server is configured to establish a baseline model of user operation behavior and identify abnormal transaction behavior by comparing the real-time collected press pattern characteristics with historical behavior patterns.

2. The secure payment system based on biometric technology as described in claim 1, characterized in that: The optical sensing unit and near-infrared imaging unit of the biometric acquisition module adopt a coaxial integrated design. By sharing an optical path, they can simultaneously capture epidermal fingerprint images and subcutaneous vein images in a single contact action. The operating wavelength of the near-infrared light source is set to the 760±5nm band, which is optimized to maximize the light absorption characteristics of hemoglobin in the vein, thereby enhancing the contrast of the vein image.

3. A secure payment system based on biometric technology as described in claim 1, characterized in that: The dynamic encryption engine performs the following operations: First, it extracts the topological structure of feature points from the fingerprint image and extracts the three-dimensional spatial coordinates of the bifurcation points of blood vessels from the vein image. These are then combined to generate a composite biometric vector. The transaction amount, the timestamp accurate to milliseconds, and the terminal device identifier are encoded into a transaction environment data block. Then, the SM4 encryption algorithm certified by the State Cryptography Administration is used, with the composite biometric vector as the encryption key, to perform encryption operations on the transaction environment data block and output a one-time valid dynamic payment token.

4. A secure payment method based on biometric technology, characterized in that, Includes the following steps: S1. In response to payment instructions, the system synchronously collects user fingerprint images, finger vein video streams, and pressure distribution time-series data through integrated sensors. S2. Analyze the physiological synchronicity between the periodic pulsation changes in blood vessel diameter in the venous video stream and the pressure fluctuations recorded by the pressure sensor to complete the in vivo verification. S3. By fusing the coordinates of fingerprint detail feature points, the spatial location of vein bifurcation points, and current transaction environment parameters, an irreversible biometric hash value is generated. S4. Embed the biometric hash value as the core verification factor into the digital signature of the payment message, and execute fund settlement after verification by the distributed ledger node.

5. A secure payment method based on biometric technology as described in claim 4, characterized in that: The specific steps of the liveness verification include: measuring the periodic change frequency of blood vessel width in a continuously acquired venous image sequence, and simultaneously detecting the fluctuation frequency of the pressure sensor output value within the corresponding time period. When both frequency values ​​are within the range of human physiological pulse characteristics and the change phases remain synchronized, it is determined that the biological characteristics originate from a living body.

6. A secure payment method based on biometric technology as described in claim 4, characterized in that: The biometric hash value generation step includes: first, encoding the spatial coordinate data of fingerprint feature points and the three-dimensional location data of vein bifurcation points into a digital feature matrix with a unified structure; then, concatenating the hash value of the terminal device hardware identifier code and the hash value of the transaction geographical location coordinates as additional factors with the digital feature matrix; and finally, using the collision-resistant Keccak-256 hash algorithm to perform multiple rounds of iterative calculations on the concatenated data set, wherein the number of iteration rounds is dynamically determined by the transaction amount value.

7. A secure payment method based on biometric technology as described in claim 4, characterized in that... It also includes the following steps: establishing a user behavior feature baseline library, continuously recording and analyzing the following behavioral parameters: the maximum pressure value during finger pressing and the time taken to reach that pressure value, the stability of the movement trajectory of the center point of the finger contact area, and the trigger probability of multiple fingers simultaneously contacting the sensor; when real-time operation parameters deviate from the historical baseline threshold, automatically activating the voice biometric secondary authentication process.

8. A payment terminal device implementing a secure payment system based on biometric technology as described in any one of claims 1-3, characterized in that: The surface of the biometric acquisition module is covered with a hardened sapphire protective layer. The bottom surface of the protective layer is etched with a micron-level lens array to eliminate ambient light reflection interference. The pressure-sensitive array is divided into independently controllable detection areas, and its detection sensitivity is dynamically adjusted according to the real-time transaction risk level. In high-risk transactions, it automatically switches to an ultra-high precision detection mode.

Images