An image recognition method for cross-domain access authentication

Abstract

The application relates to the technical field of image recognition, and discloses an image recognition method for cross-domain access authentication, which comprises the following steps: acquiring an image to be authenticated uploaded by an image acquisition node; transforming the image to be authenticated to a frequency domain, and extracting complex response signals under multiple preset directions and scales; calculating a phase consistency measure by using real and imaginary response components, and generating a phase consistency feature map; statistically analyzing local phase distribution of the phase consistency feature map, determining phase distribution entropy representing physical properties of an access medium, and accordingly reducing the weight of the phase consistency feature map; extracting geometric topological relations of structure anchor points in the feature map after the reduction, constructing an identity representation vector, and comparing the identity representation vector, the application recognizes and suppresses phase interference introduced by non-natural display media, realizes live body detection in identity authentication, and eliminates the influence of heterogeneous imaging amplitude distortion on recognition accuracy.

Description

Technical Field

[0001] This invention relates to an image recognition method for cross-domain access authentication, belonging to the field of image recognition technology. Background Technology

[0002] Currently, in distributed access authentication systems, extracting features from the raw images collected by the terminal to be authenticated and comparing them for identity verification is the mainstream technical approach. Existing technologies typically use deep neural networks to perform feature mapping in the image spatial domain, attempting to transform the pixel distribution of different sensors to a unified feature space through nonlinear transformations. However, due to the heterogeneous hardware characteristics of access terminals, there are performance differences in their sensor sensitivity and lens resolution, resulting in nonlinear amplitude distortion in the raw images. This amplitude response deviation caused by hardware makes the spatial feature extraction process dependent on massive amounts of cross-domain aligned data. When the terminal to be authenticated is in a low-light environment or under conditions with limited optical resolution, the spatial feature mapping function is prone to deviation, reducing recognition accuracy. The industry has tried to fit the cross-domain feature distribution by increasing model depth or introducing domain adversarial networks, which increases the computational load on edge devices and fails to eliminate the interference of hardware optical gain on image structure and semantics from a physical mechanism perspective.

[0003] Besides the inherent performance differences in front-end hardware, existing access control methods also exhibit logical limitations in handling the consistency of cross-domain signals. For example, Chinese invention patent CN114037457B discloses a cross-domain access authentication method for complex industrial product terminals based on identity identification. It utilizes identifier cryptography and session key negotiation mechanisms to optimize the registration process during terminal movement at the protocol level, effectively reducing resource consumption during inter-domain authentication. However, this type of control logic often presupposes the standardization and purity of the front-end acquired data, ignoring physical gain interference during the imaging process. When the signal to be authenticated is affected by sensor thermal noise, directional interference, etc., the authentication logic becomes more complex. When motion blur or geometric distortion of wide-angle lenses occurs, relying solely on the handshake logic of upper-layer protocols is insufficient to compensate for the collapse in the accuracy of the original image's physical representation. At the same time, existing solutions generally lack the ability to perceive the physical properties of the access medium, and cannot identify and block the injection of non-natural phase signals such as high-definition screen re-photographs within the authentication framework. This results in insufficient security redundancy when the system faces physical layer attacks in distributed heterogeneous environments. To address the amplitude distortion and environmental adaptability bottlenecks caused by distributed heterogeneous terminals, bypassing the unstable spatial pixel intensity distribution and extracting structurally invariant features decoupled from hardware gain characteristics from the frequency space has become a research direction for improving the reliability of cross-domain authentication.

[0004] Therefore, how to construct an image recognition mechanism that can decouple the optical characteristics of terminal hardware and has environmental adaptability to achieve identity authentication in heterogeneous terminal environments has become the technical problem to be solved by this invention. Summary of the Invention

[0005] To address the problems mentioned in the background art, the technical solution of the present invention is as follows: An image recognition method for cross-domain access authentication, comprising the following steps:

[0006] Step S1: Obtain the image to be authenticated uploaded by the image information acquisition node;

[0007] Step S2: Transform the image to be authenticated from the spatial domain to the frequency domain, and use a multi-scale logarithmic Gaussian filter bank to perform filtering operation on the frequency domain image to extract the complex frequency response signal of the image to be authenticated in multiple preset directions and multiple preset scales. The complex frequency response signal is composed of real response components and imaginary response components.

[0008] Step S3: Calculate the sum of the local energies of the real response component and the imaginary response component as the local total energy, calculate the magnitude of the sum of the response vectors of the real response component and the imaginary response component at each pixel position, determine the ratio of the magnitude to the local total energy as the phase consistency measure of the image to be authenticated at each pixel position, and generate a phase consistency feature map based on the phase consistency measure of each pixel position.

[0009] Step S4: Statistically analyze the phase angle distribution of the phase consistency feature map at each pixel position as the preset scale changes, and calculate the phase distribution entropy that characterizes the physical properties of the access medium; perform a numerical subtraction operation between the phase distribution entropy and the preset medium discrimination threshold, and map the difference obtained into a feature suppression factor for each pixel position; use the feature suppression factor to perform a weighted operation on the feature amplitude of the corresponding pixel position in the phase consistency feature map to obtain the suppressed feature map;

[0010] Step S5: Extract the set of local extreme points from the suppressed feature map as structural anchor points, construct an identity representation vector based on the geometric distribution and topological relationship of the structural anchor points, perform a matching operation between the identity representation vector and the preset registration template, and output an authentication command for the image information acquisition node based on the matching operation result.

[0011] Preferably, the calculation of phase consistency measure in step S3 further includes: obtaining the sum of component amplitudes of the real response component and the imaginary response component at each preset scale, and accumulating the component amplitude sums at each preset scale to calculate the local total energy; the magnitude of the sum of local response vectors is obtained by calculating the square root of the sum of the squares of the sum of the real response components and the sum of the squares of the sum of the imaginary response components at each preset scale, so as to characterize the edge structure of the image to be authenticated by calculating the phase resonance intensity between each preset scale.

[0012] Preferably, in step S2, the filters in the multi-scale logarithmic Gaussian filter bank are distributed in a geometric multiple in the logarithmic frequency space, and the center frequencies of each filter overlap with each other, so as to lock the invariant features of the image to be authenticated under different optical resolutions.

[0013] Preferably, the method further includes step S6: obtaining the focal length parameters of the optical imaging module in the image information acquisition node; based on the radial distance from the pixel to the image center... Calculate each preset scale Compensation phase term of the lower filter operator The formula is expressed as follows: ,in, and The preset edge distortion correction coefficients are used; the complex frequency response signal is pre-counted using a compensation phase term to correct the phase shift caused by the radial distortion of the optical imaging module.

[0014] Preferably, in step S4, the phase distribution entropy is obtained by calculating the phase angle variance of each pixel position at each preset scale, which is used to quantify the phase alignment consistency of the image to be authenticated in the frequency domain space.

[0015] Preferably, in step S4, performing weight reduction operation using the feature suppression factor includes: when the phase distribution entropy is greater than the medium discrimination threshold, determining that the value of the feature suppression factor is less than 1; multiplying the feature amplitude at the corresponding position with the feature suppression factor to weaken the periodic stripe features introduced by the non-natural display medium.

[0016] Preferably, in step S5, constructing the identity representation vector based on the geometric distribution topology of the structural anchor points includes: extracting the local phase information and coordinate information of each structural anchor point, and encoding the local phase information as a feature element according to the arrangement order of the coordinate information into an identity representation vector.

[0017] Preferably, before step S1, the method further includes: identifying the device type label of the image information acquisition node, and retrieving the corresponding preset channel noise model according to the device type label, and performing grayscale mean equalization processing on the raw image stream to be acquired.

[0018] Preferably, step S2 specifically includes: mapping the image to be authenticated from the spatial domain to the frequency domain using a fast Fourier transform; performing a point-by-point product operation between the mapped image signal and a log-Gaussian filter operator in the frequency domain; and performing an inverse Fourier transform on the result of the product operation to obtain the real and imaginary response components of the corresponding direction and scale.

[0019] Preferably, the method further includes step S7, where, when the authentication command indicates successful authentication, an authorization message containing an encrypted access token is generated and the authorization message is sent to the image information acquisition node.

[0020] Compared with the prior art, the beneficial effects of the present invention are:

[0021] 1. In image recognition, a feature alignment mechanism based on the orthogonality of physical signals is established to eliminate amplitude response deviations between heterogeneous acquisition terminals. By extracting the phase consistency measure of the image in the frequency domain, the recognition process is freed from dependence on the absolute value of pixel amplitude. Utilizing the phase resonance characteristics of image structural features in the frequency space, the hardware-related optical calibration gain terms contained in the denominator and numerator are canceled through normalized ratio operation during the calculation process, thereby decoupling amplitude response from structural semantics and obtaining native invariant feature maps that are independent of differences in terminal sensor sensitivity and lens resolution fluctuations. This eliminates the need for cross-domain authentication processes to rely on massive amounts of cross-domain alignment data for pre-training, thus eliminating feature alignment failures caused by uneven terminal hardware performance in distributed access scenarios.

[0022] 2. To achieve multi-dimensional feature reuse of intermediate variables in the frequency domain and synergistically enhance the recognition stability under extreme environments, the system utilizes the complex response vector generated by multi-scale logarithmic Gaussian transform to simultaneously perform noise floor adaptive masking and asymmetric spectral entropy directionality compensation. By extracting the local energy variance between adjacent scales, the system identifies electronic noise and effective structures. During the fusion stage, weights are allocated based on the spectral entropy of different angular directions. This enables the system to actively suppress random phase disturbances caused by sensor thermal noise in complex environments with near-zero illumination or directional motion ambiguity, and to bypass damaged physical information channels. The system reconstructs the identity representation vector based solely on the undisturbed directional components, thereby improving the feature purity of the authentication system under extreme dynamic conditions without introducing additional denoising operators.

[0023] 3. Construct an intrinsic anti-counterfeiting barrier based on cross-scale phase-locking characteristics. Identify the physical attributes of the access medium by utilizing the information redundancy difference between real physical entities and artificially reproduced media in the multi-scale frequency domain space. Calculate the phase distribution entropy by statistically analyzing the local phase angle distribution of each pixel position at all observation scales. This identifies unnatural phase abrupt changes introduced by the sub-pixel arrangement of the display device or printed dots. When the phase distribution entropy of a specific area exceeds the preset medium discrimination threshold, the system automatically performs local salience weighting on the feature map, blocking the information transmission of the counterfeit access source. This allows the identification algorithm to perform liveness detection while performing identity verification, solving the security risks brought by high-definition screen re-photographing within a single processing framework. Attached Figure Description

[0024] Figure 1 This is a flowchart of the core algorithm for cross-domain access authentication image recognition in this invention;

[0025] Figure 2 This is a schematic diagram of the hardware deployment architecture and data interaction principle of the authentication system of this invention. Detailed Implementation

[0026] The technical solutions in the embodiments of the present invention will be clearly and completely described below. 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 making objective judgments are within the scope of protection of the present invention. The following embodiments are intended to explain the present invention, and not to limit the scope of protection of the present invention.

[0027] This invention provides an image recognition method for cross-domain access authentication. By converting the image to be authenticated from the spatial domain to the frequency space, and utilizing the phase resonance characteristics of different frequency components at the visual structure, a structural feature map decoupled from the terminal hardware optical gain is extracted. Furthermore, frequency domain intermediate variables are used collaboratively to perform noise floor suppression, directional blur compensation, medium property identification, and radial distortion calibration. Finally, an identity authentication command is output based on the topological relationship of the structural anchor points. To address the nonlinear amplitude distortion problem in images acquired by heterogeneous terminals, the system acquires the image to be authenticated uploaded by the image information acquisition node, uses a Fast Fourier Transform to map the image from the spatial domain to the frequency domain, and calls a preset multi-scale logarithmic Gaussian filter bank to perform frequency domain decomposition. The multi-scale logarithmic Gaussian filter bank is geometrically distributed in the logarithmic frequency space, with overlapping center frequencies, to capture complex frequency response signals at different optical resolutions. The complex frequency response signal is composed of real response components. With the imaginary response component Composition, among which, For filter scale, In a specific engineering deployment scenario, the system receives a 640×480 resolution image to be authenticated, selects filter banks at four scales and six directions to perform point-by-point multiplication operations, and performs an inverse Fourier transform on the results. This yields a complex response vector field reflecting local frequency characteristics at each pixel location, thereby converting visual structural information from amplitude dependence to frequency distribution. Since differences in the sensitivity of different sensors are reflected in pixel amplitudes, causing a shift in recognition features across domains, the system employs a phase consistency measure to construct a structure-invariant feature map, specifically by calculating the coordinates of each pixel location. The sum of the local amplitudes is taken as the local total energy, i.e. Simultaneously, the magnitude of the sum of response vectors at each scale is calculated, and the ratio of the magnitude to the local total energy is determined as the phase consistency measure. The calculation formula is as follows: ,in, For phase consistency measures at each pixel location, For frequency weighting factors, For local amplitude, For phase deviation function, The noise threshold parameter is determined based on the energy distribution in the unstructured region. To avoid a constant with a denominator of zero, for example, a value of 0.0001, the calculation process cancels out the hardware-related optical calibration gain terms shared by the denominator and numerator through normalized ratio operations, thereby decoupling the amplitude response from the structural semantics and making the generated phase consistency feature map stable to illumination fluctuations.

[0028] Noise threshold parameter Calibration involves acquiring the image to be authenticated, selecting an unstructured reference region at the edge of the viewfinder, statistically analyzing the component amplitudes and statistical distribution characteristics of pixel positions within the reference region at various preset scales, and calculating the component amplitudes and mean values. and standard deviation According to the formula Determine the noise threshold parameter ,in To eliminate false feature points, an energy threshold is set. For the component amplitude and average value within the reference region, For component amplitude and standard deviation, To pre-set the reliability adjustment coefficient, the adjustment coefficient The numerical values ​​are selected in the range of 2.0 to 3.0. The noise floor suppression benchmark is adjusted under different illumination conditions to focus the phase consistency measurement calculation on the edges of the physical structure. To address the problem of random phase perturbations caused by sensor thermal noise in low-light environments, which lead to pseudo-structure interference in the feature map, the system adopts an adaptive noise floor masking method. The system monitors the local energy variance of the frequency domain response between adjacent scales. To identify effective features and electronic noise, a 3×3 spatial window is selected, and the dispersion of energy values ​​at each scale within this window is statistically analyzed. If the energy variance of a certain pixel... If the noise floor distribution model threshold is exceeded and the mean energy is lower than the preset energy threshold, the point is determined to be dominated by noise. In this case, the system generates a feature suppression factor and performs weighted contraction processing on the phase consistency measure at that location. The feature amplitude at the corresponding location is multiplied by an attenuation coefficient between 0 and 0.5, thereby filtering out false feature points caused by low signal-to-noise ratio without sacrificing true edge sharpness, ensuring the physical determinism of the recognition benchmark in low-light environments. To address the directional motion blur problem caused by user handheld device jitter, the system adopts a directional compensation method based on spectral entropy. The system statistically analyzes the position of each pixel in different directions... The sum of the spectral energy under the given conditions, and the spectral entropy of the signal in each direction is calculated. The direction with higher entropy value corresponds to a clear structural edge, while the direction with lower entropy value indicates frequency attenuation, i.e., blurriness. Based on the magnitude of the spectral entropy in each direction, the system uses a weighting formula to determine the weight allocation, as follows: ,in, For the corresponding direction The fusion weights; The system uses the spectral entropy of the direction to improve the stability of the recognition algorithm in dynamic interactive scenarios by increasing the weight of the clear direction in feature fusion and suppressing the phase shift interference of the blurred direction. This allows the system to bypass the damaged physical information channel and reconstruct the identity representation vector using the direction component with a lower degree of interference.

[0029] To defend against identity spoofing attacks using high-definition screens or printed materials, the system identifies the physical properties of the access medium through cross-scale phase-locking characteristics. Because display devices introduce specific periodic sampling patterns when reconstructing images, their phase distribution in the multi-scale frequency domain exhibits abrupt phase changes. The system extracts the local phase angles of each pixel location at different scales. The phase distribution entropy of a specific region is calculated. If the phase distribution entropy of a specific region exceeds a preset medium discrimination threshold (in this embodiment, the threshold can be set to 0.85), then the region is determined to have a medium injection risk. At this time, the weight of the region in the phase consistency feature map is forcibly reset to zero to achieve the liveness detection function. The feature suppression factor of each pixel position is determined by calculating the difference between the phase distribution entropy and the medium discrimination threshold. Characterizing the strength of non-natural properties of the access medium using the formula Determine the characteristic inhibition factor ,in These are the weighting coefficients applied to the corresponding positions of the phase consistency feature map. The phase distribution entropy minus the difference in medium discrimination threshold values. The preset suppression strength gain coefficient, gain coefficient The value range is set to 1.2 to 3.5 based on the subpixel arrangement frequency of mainstream display terminals in 2025. This causes the response amplitude of the region with high entropy value forged signal to be attenuated to below the background noise level in the feature map, thus cutting off the propagation path of non-natural signals at the physical feature level.

[0030] To address the issue of radial phase distortion at image edges caused by wide-angle lenses used in distributed terminals, the system implements coordinate-driven nonlinear phase pre-counting to calculate the coordinates of each pixel position relative to the image center. Normalized radial distance The phase compensation offset at each scale is determined according to the correction formula, which is as follows: ,in, This is the phase compensation offset. Normalized radial distance; For filter scale; and To correct edge distortion based on pre-calibrated edge distortion coefficients according to lens focal length parameters, the system performs multi-scale log-Gaussian filtering, adjusting the phase compensation offset corresponding to each pixel position. As a phase-shifting operator acting on a complex frequency response signal, through a complex rotation factor The response components of the filter operator are multiplied point by point to perform a phase delay compensation during the frequency domain decomposition stage, so that the frequency components in the distorted region are physically aligned in the spatial position after the inverse transformation.

[0031] After obtaining the phase consistency feature map, the system extracts a set of local extrema points from the suppressed feature map and identifies this set as structural anchor points. By recording the local phase and coordinate information of each structural anchor point, the system converts the information into an identity representation vector according to a preset encoding order. The geometric distribution topology of the structural anchor points and the construction of the identity representation vector adopt a polar coordinate system concatenated encoding method. By taking the geometric center of the image to be authenticated as the origin, the radial distance of each structural anchor point is extracted. Polar angle and local phase value According to polar angle From 0 to Sort sequentially and extract the local phase values ​​corresponding to each point after sorting. Performing cascade operations generates a fixed-length original feature sequence, and using a normalization function to perform amplitude scaling to generate an identity representation vector. ,in To match the feature vector, the encoding order is determined to eliminate the interference of image rotation or translation on the consistency of feature expression, so that the matching operation is anchored to the physical feature dimension. The vector is matched with the registration template pre-stored in the server. If the matching score exceeds the authentication threshold, an authorization message containing an encrypted access token is generated and sent to the image information acquisition node. The whole process is implemented through a determined physical feature model, reducing the computational load on the gateway side.

[0032] Example 1: In a nighttime access authentication scenario in a logistics park, due to ambient light intensity below 5 lux and nonlinear differences in the amplitude response curves of different access terminal photosensitive units, traditional spatial pixel features fail to recognize each other during cross-domain comparison due to contrast loss. In this case, the image recognition method acquires the image to be authenticated through image information acquisition nodes, maps the image from the spatial domain to the frequency domain, and uses a multi-scale logarithmic Gaussian filter bank to extract complex frequency response signals at four scales and six directions, calculating the coordinates of each pixel. The sum of local amplitudes is taken as the local total energy. At the same time, the magnitude of the sum of response vectors at each scale is calculated, and the ratio of the magnitude to the local total energy is determined as the phase consistency measure. The calculation formula is as follows: ,in, A measure of phase consistency at each pixel location; Frequency weighting factor; This refers to the local amplitude. The phase deviation function is defined as a fixed calibration vector pre-stored in a static buffer in the calculation process. Its value originates from the dark current phase shift sampling of 4096 pixels of the sensor in a dark environment during the initial power-on of the system. The value range is calibrated between 0.005 rad and 0.025 rad, serving as a static denominator correction term for the phase consistency measure. The phase pre-adjustment process in step S6 performs secondary real-time fine-tuning for the dynamic distortion caused by the shift of the lens's physical center of gravity. The two processes work together to lock the physical phase coordinates through linear accumulation. This is the noise threshold parameter; The value is a constant of 0.0001. This calculation process cancels out the hardware-related optical calibration gain term through normalized ratio operation, thereby decoupling the amplitude response from the structural semantics.

[0033] Under conditions where sensor thermal noise and directional motion ambiguity are superimposed, image recognition methods monitor the local energy variance of the frequency domain response between adjacent scales. To identify electronic noise, a 3×3 spatial window is selected, and the dispersion of energy values ​​at each scale within this window is statistically analyzed. If the energy variance of a certain pixel is... If the value exceeds a preset threshold and its mean energy is below a preset threshold, then the point is determined to be dominated by noise. In this case, a feature suppression factor is generated to perform weighted shrinkage processing on the phase consistency measure, and the position of each pixel in different directions is statistically analyzed. Sum the spectral energy and calculate the spectral entropy of the signal in each direction. The weight allocation is determined based on the magnitude of the spectral entropy in each direction using a weighting formula, as follows: ,in, For fusion weights; To address the spectral entropy, by increasing the weight of the sharp direction and suppressing phase shift interference in the blurred direction, the image recognition method bypasses the damaged physical information channel. It reconstructs the identity representation vector using the less disturbed direction components. To address the radial phase distortion caused by wide-angle lenses, the image recognition method implements coordinate-driven nonlinear phase pre-counterbalancing, calculating the normalized radial distance of each pixel position relative to the image center coordinates. The phase compensation offset at each scale is determined according to the correction formula. ,in, This is the phase compensation offset. For normalized radial distance, For filter scale, and As the edge distortion correction coefficient, phase shift correction is performed on the complex response vector using phase compensation offset, so that the frequency components of the distorted region are re-phase aligned at their corresponding geometric positions, eliminating phase tailing and ensuring consistent recognition accuracy for the user across the entire viewfinder. After obtaining the phase consistency feature map, the image recognition method extracts structural anchor points from the feature map, records their local phase and coordinate information, converts them into identity representation vectors according to a preset encoding order, and performs cosine similarity matching with the pre-stored registration template on the server. When the matching score exceeds the authentication threshold of 0.85, an authorization message containing an encrypted access token is generated and sent to the image information acquisition node. The entire process is implemented through a defined phase domain physical feature model, which makes the identity authentication result inherently stable against illumination fluctuations, sensor noise, and lens distortion.

[0034] Example 2: In a nighttime industrial logistics park access scenario with illuminance ranging from 1 lux to 10 lux, the nonlinear differences in the photosensitivity of the imaging units, which include both fixed-focus monitoring units and mobile smart terminals, cause recognition shifts in image signals during cross-domain comparisons. To verify the effectiveness of the mechanism, an experiment was conducted on a distributed computing verification platform containing two image acquisition nodes with different dynamic range characteristics. During the experiment, Gaussian white noise with a signal-to-noise ratio of 25 dB was injected into the original image data to simulate sensor thermal noise disturbance. The sampling resolution was set to 1280×720, and the filter bank scale n was set to 4 with directional... Setting the value to 6, taking the 5 lux condition as an example, the average contrast of the original image data in the spatial domain is less than 0.2 and is accompanied by pixel-level particle interference. At this time, the feature map output by the control group shows edge collapse, and its recognition accuracy is measured to be 62.4% on average. The sample group of this invention maps the original image containing 25dB noise to the frequency domain by performing a fast Fourier transform, and the system calculates the local energy variance between adjacent scales. Measurement data shows that at the noise-dominant point, its energy variance The energy variance is between 1.28 and 1.56, while the energy variance at the structural edges is... The value is below 0.35. Based on this physical characteristic difference, the system uses an adaptive noise floor mask to suppress the phase weight of the noise region by 85%. In the suppressed complex response field, the spectral entropy of the clear edge direction is... The measured value was 4.82, while the directional spectral entropy affected by motion ambiguity decreased to 2.15. This was achieved by using the formula... Determine the weight allocation and clarify the direction to obtain the fusion weights. The peak signal-to-noise ratio of the processed phase consistency feature map under 5 lux illumination was increased to 0.72, from 18.4 dB in the original data to 28.6 dB, confirming that the adaptive noise floor mask provides a high signal-to-noise ratio input prerequisite for directional compensation.

[0035] For key parameter filter scale The value boundary is determined by setting multiple gradient comparison sample groups in the experiment. When the median value is 4, the system achieves a recognition accuracy of 91.2% at 1 lux illumination, while when... When the value is lowered to 2, the recognition accuracy drops to 76.5%. Increasing the threshold to 6 only improved the recognition accuracy to 91.8%, but resulted in a 142ms increase in system computation latency, indicating that the system... The system reaches a performance saturation point around 4. Further verification using a partially missing control group showed that after removing the adaptive noise floor mask, the false recognition rate in a 1-lux environment increased from 0.8% in the original sample to 12.4%. This performance inflection point confirms the causal relationship between phase resonance characteristics and noise suppression mechanisms under low illumination. To verify the correction effectiveness of radial phase distortion, the system addresses the edge phase trailing phenomenon caused by wide-angle lenses using a correction formula. Perform phase shift compensation at normalized radial distance For the viewfinder edge region with a value of 0.9, the measured value of the uncompensated complex response phase deviation is 0.82 rad. After offsetting with correction coefficients, the residual phase deviation is reduced to less than 0.05 rad. As a result, the topological consistency measure of the extracted structural anchor points in coordinate space is restored from 0.64 to 0.92. Finally, the consistency deviation of the matching score generated based on cosine similarity between heterogeneous terminals remains within 2.8%.

[0036] Example 3: This example combines Figures 1 to 2 This describes an image recognition method for cross-domain access authentication, such as... Figure 1As shown, step S1 acquires the image to be authenticated uploaded by the image information acquisition node. Step S2 transforms the image to be authenticated from the spatial domain to the frequency domain, performs filtering operation using a multi-scale logarithmic Gaussian filter bank, and extracts complex frequency response signals containing real and imaginary components under multiple preset directions and multiple preset scales. Then, step S3 calculates the magnitude of the sum of local energies of the real and imaginary response components, i.e., the sum of local total energy and response vectors. The ratio of the magnitude to the local total energy determines the phase consistency measure of each pixel position, generating a phase consistency feature map. Next, step S4 statistically analyzes the phase angle distribution of the phase consistency feature map as the scale changes, calculates the phase distribution entropy representing the physical properties of the access medium, and generates a feature suppression factor based on the difference between the phase distribution entropy and the medium discrimination threshold. A weighted operation is performed on the feature amplitude to obtain the suppressed feature map. Finally, step S5 extracts the set of local extreme points from the suppressed feature map as structural anchor points, constructs an identity representation vector based on its geometric distribution topology, performs a matching operation with the registration template, and outputs an authentication command based on the result.

[0037] like Figure 2 As shown, the system front end is deployed with image information acquisition nodes, which include access terminals, fixed-focus monitoring units, or mobile smart terminals. The internal hardware integrates optical imaging modules, sensors, wide-angle lenses, and preset channel noise models. It is responsible for uploading images to be authenticated to the gateway and receiving authorization messages containing encrypted access tokens. The core of the system's data processing is the gateway-side embedded device, which performs identity authentication tasks. Its internal logic components include a multi-scale log-Gaussian filter bank and an identity authentication module. The identity authentication module specifically performs the operation logic of phase consistency measure calculation, phase distribution entropy calculation, liveness detection, and identity representation vector construction. At the same time, the device is equipped with a storage unit to store edge distortion correction coefficients, filter center frequency, and noise threshold parameters. The gateway-side embedded device connects to the server that provides backend data support through a matching operation interaction interface and is mainly responsible for maintaining the pre-stored registration templates.

[0038] Example 4: In an access authentication risk test environment containing a high-definition display terminal with a resolution of no less than 1920×1080 and paper printing media, the image recognition method acquires the image to be authenticated through the image information acquisition node. To determine the noise suppression boundary in the phase domain decomposition stage, the system statistically analyzes the local total energy of the unstructured components within a 32×32 pixel reference area at the image edge. Calculate the energy mean of the unstructured components. With energy standard deviation And according to the noise threshold formula Determine the noise threshold parameter ,in, This is the noise threshold parameter; The average energy value; Energy standard deviation; The gain adjustment coefficient is set to 2.5 in this embodiment, thereby achieving adaptive calibration of the phase feature extraction process based on the current noise floor level of the imaging unit. To identify the physical properties of the access medium, the image recognition method executes the phase distribution entropy extraction procedure, specifically extracting the local phase angle of each pixel position at different scales. Mapping to a phase histogram containing 16 quantization intervals, and counting the frequency of occurrence of each quantization interval. Calculate the phase distribution entropy of each pixel region, where, For the first The phase distribution probability of each quantization interval is used to determine the risk of non-natural display medium injection when the phase distribution entropy in a continuous 8×8 pixel area exceeds the medium discrimination threshold of 0.85. At this time, the weight of this area in the phase consistency feature map is reset to 0 by the feature suppression factor.

[0039] After extracting the set of structural anchor points, the system initiates a 7x7 pixel sliding detection window in the suppressed feature map. Only when the phase consistency value of the center point of the window is higher than 1.25 times the average value of all pixels in the window, and the absolute value of the center point exceeds the hard threshold of 0.15, is it identified as a structural anchor point. This filters out false feature points caused by low-contrast textures. When reading the local phase information of each structural anchor point, the system uses an 8-direction hierarchical quantization method to map the phase angle of each feature point to a preset quantization interval index set, and coordinates it with its normalized radius coordinates and angular coordinates in the polar coordinate system to form a feature tuple. Each tuple is arranged in descending order of preset feature point degree to form a fixed-length identity representation vector, which is input to the matching module. The system calculates the Hamming distance between the identity representation vector and the pre-stored registration template on the server. If the Hamming distance is less than the preset authentication tolerance, an authorization message containing an encrypted access token is generated and sent to the image information acquisition node. This process realizes the closed loop of identity authentication results at the physical feature level through quantization parameter calibration and feature sequence uniqueness mapping.

[0040] Example 5: In a distributed access scenario containing imaging units of various sizes, the cutoff frequency distribution of the original image data shifts with the change in photosensitive pixel size. Before deploying the image information acquisition nodes, the system performs parameter initialization of a multi-scale logarithmic Gaussian filter bank. By inputting a preset sinusoidal pulse texture signal to the image acquisition nodes, it monitors the parameters of each scale. The amplitude concentration of the lower complex response signal is determined based on the real part of the response component. With the imaginary response component Based on the energy distribution pattern, the center frequency of the filter that makes the edge responsivity of the structure high is selected, and the center frequency of the filter is written into the storage unit as the calibration reference for the phase consistency energy extraction stage.

[0041] When the system faces electronic noise floor interference caused by sensor manufacturing tolerances, the system acquires reference bias data from image information acquisition nodes in a dark environment, statistically analyzes the complex response distribution characteristics of the reference bias data in the frequency domain, and calculates the average energy under unstructured background. With energy standard deviation The gain adjustment coefficient is determined based on the security level of the current application scenario. The value is used to generate the noise threshold parameter. It is also injected into the identity authentication module on the gateway side, so that the system can establish a feature noise suppression baseline that matches the hardware attributes of the image information acquisition node during the process of performing hash mapping and Hamming distance calculation.

[0042] Example 6: In a distributed deployment scenario involving multiple wide-angle imaging units, the system executes a calibration procedure for radial phase distortion parameters, acquires a checkerboard calibration target image located at a preset object distance, and calculates the normalized radial distance of each corner coordinate relative to the image center coordinate. Simultaneously, the local phase response values ​​of each corner point in the multi-scale filtered space are extracted, the cross-scale phase deviation observations are recorded, and a normalized radial distance is constructed. Filter Scale and phase compensation offset The observation matrix is ​​then subjected to surface fitting using the least squares method to determine the edge distortion correction coefficients. and The numerical solution is obtained and written into the storage unit. This calibration process provides operator input parameters for performing coordinate-driven nonlinear phase pre-pinch.

[0043] During the initialization phase of the system's identity authentication module, the determination of the authentication threshold and media discrimination threshold follows a statistical procedure based on the preset security level. The system selects a verification sample set containing authorized and unauthorized identities for matching operations, statistically analyzes the false recognition rate distribution under different matching score intervals, and selects the score value with a false recognition rate of less than 0.01% to determine the identity authentication threshold. For the sampling features generated by the display medium, the system tests the phase distribution entropy measurement value under different refresh rates, and determines 1.2 times the distribution mean as the media discrimination threshold. This debugging procedure converts statistical laws into logical judgment thresholds, enabling the image recognition method to establish a discrimination standard that matches the physical conditions when facing different imaging hardware and access environments.

[0044] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0045] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An image recognition method for cross-domain access authentication, characterized in that, Includes the following steps: Step S1: Obtain the image to be authenticated uploaded by the image information acquisition node; Step S2: Transform the image to be authenticated from the spatial domain to the frequency domain, and use a multi-scale logarithmic Gaussian filter bank to perform filtering operation on the frequency domain image to extract the complex frequency response signal of the image to be authenticated in multiple preset directions and multiple preset scales. The complex frequency response signal is composed of real response components and imaginary response components. Step S3: Calculate the sum of the local energies of the real response component and the imaginary response component as the local total energy, calculate the magnitude of the sum of the response vectors of the real response component and the imaginary response component at each pixel position, determine the ratio of the magnitude to the local total energy as the phase consistency measure of the image to be authenticated at each pixel position, and generate a phase consistency feature map based on the phase consistency measure of each pixel position. Step S4: Statistically analyze the phase angle distribution of the phase consistency feature map at each pixel position as the preset scale changes, and calculate the phase distribution entropy that characterizes the physical properties of the access medium; perform a numerical subtraction operation between the phase distribution entropy and the preset medium discrimination threshold, and map the difference obtained into a feature suppression factor for each pixel position. By using a feature suppression factor to perform a weighting operation on the feature amplitude at the corresponding pixel position in the phase consistency feature map, a suppressed feature map is obtained. Step S5: Extract the set of local extreme points from the suppressed feature map as structural anchor points, construct an identity representation vector based on the geometric distribution and topological relationship of the structural anchor points, perform a matching operation between the identity representation vector and the preset registration template, and output an authentication command for the image information acquisition node based on the matching operation result.

2. The image recognition method for cross-domain access authentication according to claim 1, characterized in that, Step S3, which calculates the phase consistency measure, also includes: obtaining the sum of the component amplitudes of the real and imaginary response components at each preset scale, and summing the component amplitudes at each preset scale to obtain the local total energy; the magnitude of the sum of local response vectors is obtained by calculating the square root of the sum of the squares of the real and imaginary response components at each preset scale, so as to characterize the edge structure of the image to be authenticated by calculating the phase resonance intensity between each preset scale.

3. The image recognition method for cross-domain access authentication according to claim 1, characterized in that, In step S2, the filters in the multi-scale logarithmic Gaussian filter bank are distributed in a geometric multiple in the logarithmic frequency space, and the center frequencies of each filter overlap with each other, so as to lock the invariant features of the image to be authenticated under different optical resolutions.

4. The image recognition method for cross-domain access authentication according to claim 1, characterized in that, The method also includes step S6: obtaining the focal length parameters of the optical imaging module in the image information acquisition node; based on the radial distance from the pixel to the image center... Calculate each preset scale Compensation phase term of the lower filter operator The formula is expressed as follows: ,in, and The preset edge distortion correction coefficients are used; the complex frequency response signal is pre-counted using a compensation phase term to correct the phase shift caused by the radial distortion of the optical imaging module.

5. The image recognition method for cross-domain access authentication according to claim 1, characterized in that, In step S4, the phase distribution entropy is obtained by calculating the phase angle variance of each pixel position at each preset scale, which is used to quantify the phase alignment consistency of the image to be authenticated in the frequency domain space.

6. The image recognition method for cross-domain access authentication according to claim 1, characterized in that, In step S4, the weighting operation using the feature suppression factor includes: when the phase distribution entropy is greater than the medium discrimination threshold, determining that the value of the feature suppression factor is less than 1; multiplying the feature amplitude at the corresponding position with the feature suppression factor to weaken the periodic stripe features introduced by the non-natural display medium.

7. The image recognition method for cross-domain access authentication according to claim 1, characterized in that, In step S5, constructing the identity representation vector based on the geometric distribution topology of the structural anchor points includes: extracting the local phase information and coordinate information of each structural anchor point, and encoding the local phase information as a feature element according to the arrangement order of the coordinate information into an identity representation vector.

8. The image recognition method for cross-domain access authentication according to claim 1, characterized in that, Before step S1, the method further includes: identifying the device type label of the image information acquisition node, and retrieving the corresponding preset channel noise model according to the device type label, and performing grayscale mean equalization processing on the raw image stream to be acquired.

9. The image recognition method for cross-domain access authentication according to claim 1, characterized in that, Step S2 specifically includes: mapping the image to be authenticated from the spatial domain to the frequency domain using a fast Fourier transform; performing a point-by-point product operation between the mapped image signal and the log-Gaussian filter operator in the frequency domain; and performing an inverse Fourier transform on the result of the product operation to obtain the real and imaginary response components for the corresponding direction and scale.

10. The image recognition method for cross-domain access authentication according to claim 1, characterized in that, The method also includes step S7, which generates an authorization message containing an encrypted access token when the authentication command indicates successful authentication, and sends the authorization message to the image information acquisition node.

Images