High-frequency-based palm living body detection method, system and equipment and storage medium

Abstract

The invention discloses a high-frequency-based palm living body detection method, system and device and a storage medium. The method comprises the steps that S1, a palm image is acquired; s2, Fourier transform is carried out on the palm image to obtain frequency spectrum information, and a high-frequency part is extracted; s3, performing Fourier inverse transformation on the high-frequency part to obtain a high-frequency palm image; and S4, carrying out living body recognition on the high-frequency palm image by using a living body recognition model. According to the invention, the palm can be accurately subjected to living body recognition.

Description

Background Art

[0002] With the rapid development of information technology, biometric technology has been widely used in many fields as a secure and reliable means of identity authentication. Among them, palm recognition has attracted much attention due to its unique advantages. The palm has rich texture and structural information, providing a variety of features for identity recognition, such as palm prints and palm veins. However, traditional palm recognition technology faces a severe challenge in terms of security, that is, the threat of forgery attacks. Forgery attacks include using non-living means such as paper-printed palm images, silicone palm molds, or palm images displayed on the screen. These attack methods may successfully bypass traditional palm recognition systems and cause serious security vulnerabilities. Therefore, ensuring that the palm recognition system can accurately distinguish between live palms and non-living samples has become a key issue that needs to be urgently solved in the field of biometrics.

[0003] In the field of signal processing, Fourier transform is a powerful tool that can convert time domain or spatial domain signals into frequency domain signals, thereby revealing the frequency composition information of the signal. For images, Fourier transform converts images from the spatial domain to the frequency domain, which can show the distribution of different frequency components in the image. In an image, the low-frequency part usually contains the main outline and basic structural information of the image, while the high-frequency part contains the detailed information of the image, such as texture, edges, etc. The study found that living palms and non-living palms have distinguishable features in the high-frequency part, which provides new ideas for developing safer palm liveness detection methods.

[0004] At present, a variety of palm liveness detection methods have been proposed, which can be mainly divided into the following categories:

[0005] Physiologically based methods:

[0006] Such methods usually rely on the physiological characteristics of the palm, such as measuring the palm's temperature, blood flow, pulse, etc. For example, infrared thermal imaging technology can be used to detect the palm's temperature distribution and blood flow to determine whether the palm is alive. However, this method requires additional special sensors, which increases the cost and complexity of the system, and its performance is easily affected by factors such as ambient temperature and measurement distance.

[0007] Behavior-based methods:

[0008] Observe the dynamic behavior of the palm, such as the movement of fingers, the change of palm pressure, etc. The dynamic characteristics of these actions are analyzed by the user performing some preset actions, such as palm translation, rotation or finger bending. However, this type of method requires the active cooperation of the user and may cause inconvenience to the user. At the same time, in some complex environments, the recognition accuracy of the action may decrease, affecting the reliability of the system.

[0009] The disclosure of the above background art content is only used to assist in understanding the inventive concept and technical solution of the present invention, and it does not necessarily belong to the prior art of this patent application. Without clear evidence showing that the above content was publicly available on the filing date of this patent application, the above background art should not be used to evaluate the novelty and inventiveness of this application. Summary of the Invention

[0010] To this end, the present invention provides a high-frequency-based palm liveness detection solution to meet the need for palm liveness detection in various application scenarios.

[0011] In a first aspect, the present invention provides a high-frequency-based palm liveness detection method, which is characterized by including:

[0012] Step S1: Obtain a palm image;

[0013] Step S2: Perform a Fourier transform on the palm image to obtain spectral information and extract the high-frequency part;

[0014] Step S3: Perform an inverse Fourier transform on the high-frequency part to obtain a high-frequency palm image;

[0015] Step S4: Use a liveness recognition model to perform liveness recognition on the high-frequency palm image.

[0016] Optionally, in the high-frequency-based palm liveness detection method, the palm image obtained in step S1 is an infrared image.

[0017] Optionally, in the high-frequency-based palm liveness detection method, step S2 includes:

[0018] Step S21: Perform a two-dimensional Fourier transform on the palm image to obtain a complex matrix;

[0019] Step S22: Calculate the spectral information;

[0020] Step S23: Determine and separate the high-frequency part from the spectral information according to a threshold.

[0021] Optionally, in the high-frequency-based palm liveness detection method, step S3 includes:

[0022] Step S31: Reconstruct a complex matrix according to the high-frequency part;

[0023] Step S32: Apply an inverse Fourier transform algorithm to convert the high-frequency part in the frequency domain back to the spatial domain;

[0024] Step S33: Process the result in the spatial domain to obtain a high-frequency palm image.

[0025] Optionally, in the described high-frequency-based palm liveness detection method, step S4 includes:

[0026] Step S41: Extract features from the high-frequency palm image to obtain palm features;

[0027] Step S42: Input the palm features into the liveness recognition model;

[0028] Step S43: Obtain the liveness recognition result.

[0029] Optionally, in the described high-frequency-based palm liveness detection method, when the liveness recognition model is trained, the palm image is used for supervision.

[0030] Optionally, in the described high-frequency-based palm liveness detection method, when the liveness recognition model is trained, by setting different high-frequency starting values, different high-frequency parts are obtained to improve the adaptability of the liveness recognition model.

[0031] In a second aspect, the present invention provides a high-frequency-based palm liveness detection system for implementing the high-frequency-based palm liveness detection method described in any one of the foregoing, which includes:

[0032] An acquisition module for acquiring a palm image;

[0033] An extraction module for performing Fourier transform on the palm image to obtain spectral information and extracting the high-frequency part;

[0034] An inverse transform module for performing inverse Fourier transform on the high-frequency part to obtain a high-frequency palm image;

[0035] A recognition module for performing liveness recognition on the high-frequency palm image using a liveness recognition model.

[0036] In a third aspect, the present invention provides a high-frequency-based palm liveness detection device, which includes:

[0037] A processor;

[0038] A memory in which executable instructions of the processor are stored;

[0039] Wherein, the processor is configured to execute the steps of the high-frequency-based palm liveness detection method described in any one of the foregoing by executing the executable instructions.

[0040] In a fourth aspect, the present invention provides a computer-readable storage medium for storing a program, which is characterized in that when the program is executed, the steps of the high-frequency-based palm liveness detection method described in any one of the foregoing are implemented.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] By performing Fourier transform on the palm image, the present invention converts it from the spatial domain to the frequency domain, enabling the analysis of the palm image from a brand-new perspective. This conversion can decompose the information in the image into components of different frequencies, revealing features that are not easily perceptible in the spatial domain. The high-frequency part usually contains detailed information in the image, such as texture, edges, and minute variations, and these details are of great significance for distinguishing between a living palm and a non-living palm. For example, due to physiological characteristics such as blood circulation and muscle activity in a living palm, the surface texture will exhibit some subtle dynamic changes, which may be reflected in the spectral information of the high-frequency part. In contrast, non-living samples such as printed palm images and silicone palms, because their textures are static, will have different characteristics in the high-frequency part from those of a living palm. Therefore, potential distinguishing features can be excavated through Fourier transform.

[0043] The present invention utilizes the characteristics of Fourier transform to endow the method with a certain resistance to some common image noises and interferences. In the frequency domain, different types of noises often have specific frequency distributions. By reasonably setting the frequency extraction range, some noises can be effectively filtered out, making the subsequent extracted high-frequency part more representative. Moreover, Fourier transform is a mature mathematical tool with a good theoretical foundation, and its results are relatively stable, being unaffected by geometric transformations such as image translation and rotation, enhancing the robustness of the method in complex environments.

[0044] The present invention performs inverse Fourier transform on the high-frequency part to obtain a high-frequency palm image, which highlights the detailed features of the palm. In this way, key features such as the texture and edges of the palm are focused on, avoiding redundant information that may be brought by the low-frequency part, enabling the subsequent living body recognition model to more centrally learn and analyze the key features, and improving the pertinence and accuracy of recognition. Compared with directly using the original palm image for recognition, the high-frequency palm image reduces the interference of the background, general outline, etc. on the recognition result, being more conducive to extracting unique features for distinguishing between a living body and a non-living body. Especially for some camouflage means, such as printed palm images in different backgrounds or non-living palms with partial occlusion, this processing method can better highlight the differences. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on the provided drawings. By reading the detailed description of the non-limiting embodiments with reference to the following drawings, other features, objectives, and advantages of the present invention will become more obvious:

[0046] Figure 1 It is a flowchart of the steps of a high-frequency-based palm liveness detection method in an embodiment of the present invention;

[0047] Figure 2 It is a schematic diagram of a high-frequency image in an embodiment of the present invention;

[0048] Figure 3 It is a schematic diagram of a frequency spectrum distribution in an embodiment of the present invention;

[0049] Figure 4 It is a flowchart of the steps of extracting the high-frequency part in an embodiment of the present invention;

[0050] Figure 5 It is a flowchart of the steps of obtaining a high-frequency palm image in an embodiment of the present invention;

[0051] Figure 6 It is a flowchart of the steps of performing liveness recognition on a high-frequency palm image in an embodiment of the present invention;

[0052] Figure 7 It is a schematic diagram of the structure of a high-frequency-based palm liveness detection system in an embodiment of the present invention;

[0053] Figure 8 It is a schematic diagram of the structure of a high-frequency-based palm liveness detection device in an embodiment of the present invention; and

[0054] Figure 9 It is a schematic diagram of the structure of a computer-readable storage medium in an embodiment of the present invention. Detailed Embodiments

[0055] The present invention will be described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but do not limit the present invention in any form. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made. These all belong to the protection scope of the present invention.

[0056] In the description, claims, and the above-mentioned drawings of the present invention, terms such as "first", "second", "third", "fourth", etc. (if any) are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present invention described herein, for example, can be implemented in an order other than those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0057] A high-frequency-based palm liveness detection method provided by an embodiment of the present invention aims to solve the problems existing in the prior art.

[0058] The following uses specific embodiments to detail the technical solutions of the present invention and how the technical solutions of the present application solve the above technical problems. These several specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The following will describe the embodiments of the present invention in conjunction with the drawings.

[0059] The present invention obtains a palm image, performs a Fourier transform on it to obtain spectral information and extracts the high-frequency part, then performs an inverse Fourier transform to obtain a high-frequency palm image, and finally uses a liveness recognition model to complete the liveness recognition of the palm based on the high-frequency palm image, having the advantages of general equipment, anti-interference, effective feature extraction, strong learning ability, scalability, high security, and good reliability.

[0060] Figure 1 It is a flowchart of the steps of a high-frequency-based palm liveness detection method in an embodiment of the present invention. As Figure 1 shown, the steps of a high-frequency-based palm liveness detection method in an embodiment of the present invention include:

[0061] Step S1: Obtain a palm image.

[0062] In this step, various devices with image acquisition functions can be used, such as ordinary digital cameras, smartphone cameras, dedicated industrial cameras, or security cameras, etc. These devices should have sufficient resolution and image quality to ensure that the detailed features of the palm can be clearly captured. The palm image can be directly acquired or obtained from the network or other channels.

[0063] Image acquisition process:

[0064] Guide the user to place the palm within the visible range of the device, ensuring that the palm is in a proper position and posture. Usually, it is required that the palm is naturally extended and the fingers are separated to collect complete palm information.

[0065] Trigger the image acquisition operation, which can be a manual operation (such as pressing the shutter button) or an automatic trigger (for example, timed photography set by the program) to obtain the original image of the palm.

[0066] The original image may be a color image, and its storage format is usually a common image format, such as JPEG, PNG, etc. The information it contains is represented in the form of a pixel matrix, and each pixel point may have color information of three RGB channels (for color images) or grayscale information (for grayscale images).

[0067] Image preprocessing:

[0068] Perform some preprocessing operations on the original image, such as converting the color image to a grayscale image, which can be achieved by the weighted average method for subsequent Fourier transform processing.

[0069] The image can also be cropped to remove unnecessary background parts, making the palm occupy the main part in the image and reducing interference information.

[0070] Preferably, to improve the quality of subsequent processing, the palm image obtained in this step is an infrared image. For the palm, the RGB image obtains the surface texture information of the palm, while the infrared image can obtain the palm vein information. For a polarized infrared camera, a clearer palm vein image can also be obtained. These images can all be used in this embodiment.

[0071] Step S2: Perform a Fourier transform on the palm image to obtain spectral information and extract the high-frequency part.

[0072] In this step, the Fourier transform is used to convert the image from the spatial domain to the frequency domain and extract the high-frequency part therein.

[0073] Fourier transform principle:

[0074] Apply the two-dimensional discrete Fourier transform (2DDFT) or two-dimensional fast Fourier transform (2DFFT) to the preprocessed palm image (usually a two-dimensional matrix). For the discrete image signal f(x,y), its two-dimensional discrete Fourier transform formula is:

[0075]

[0076] where M and N are the dimensions of the image, u and υ are the coordinates in the frequency domain, and j is the imaginary unit.

[0077] The Fourier transform decomposes an image into a superposition of sine and cosine waves of different frequencies, resulting in a complex matrix F(u, υ), which contains the amplitude and phase information of the image at different frequencies.

[0078] Spectrum information calculation:

[0079] Calculate the amplitude spectrum of the spectrum using the formula Obtain the amplitude information of different frequency components in the frequency domain, which represents the contribution degree of each frequency component to the image.

[0080] When visualizing the amplitude spectrum, usually the zero-frequency component (DC component) is moved to the center of the spectrum to better observe the distribution of low-frequency and high-frequency information. This can be achieved by performing a translation operation on the frequency matrix.

[0081] High-frequency part extraction:

[0082] Define the standard for the high-frequency part. Generally speaking, the high-frequency part refers to the part farther from the center of the spectrum, because in the spectrum, the low-frequency part is usually concentrated in the center, while the high-frequency part is distributed around.

[0083] Set a frequency threshold. For example, determine a radius r from the center of the spectrum through experiments or according to experience. Consider the part with a distance greater than r as the high-frequency part, and set the frequency components below this threshold to zero or perform low-weight processing, so as to extract the spectrum information of the high-frequency part.

[0084] Figure 2 Images of different spectra of the palm are shown. In Figure 2 ,"Orig" represents the original image, "lowfrerquency" represents the low-spectrum image, and "high frequency" represents the high-frequency image. It can be seen from the figure that the high-frequency part obviously has characteristics different from those of the low-spectrum part and the original image.

[0085] Figure 3 For the spectrum distribution. Figure 3 In,"Low Frequency" represents the spectrum distribution of the low-frequency part, and "HighFrequency" represents the spectrum distribution of the high-frequency part. It can be seen from Figure 3 that the distributions of the two are significantly different and can be used for live body recognition.

[0086] Step S3: Perform an inverse Fourier transform on the high-frequency part to obtain a high-frequency palm image.

[0087] In this step, the high-frequency part in the frequency domain is converted back to the spatial domain.

[0088] Principle of inverse Fourier transform:

[0089] Apply the two-dimensional inverse discrete Fourier transform (2DIDFT) or two-dimensional inverse fast Fourier transform (2DIFFT) to the extracted high-frequency part spectrum (complex matrix), and its formula is:

[0090]

[0091] where F(u, υ) is the spectrum information of the high-frequency part.

[0092] The inverse transform converts the information in the frequency domain back to the spatial domain, and the resulting result is still a complex matrix.

[0093] Process the inverse transform result:

[0094] For the complex matrix after the inverse transform, usually take its magnitude as the final image information because the magnitude represents the intensity of the signal. The formula can be used to obtain the final image pixel value.

[0095] The obtained image is the high-frequency palm image, which emphasizes the high-frequency details in the original palm image, such as texture and edge information, while the information in the low-frequency part (such as the general outline) is weakened, which is more conducive to subsequent live body recognition steps because there may be distinguishable features in the details such as texture between live and non-live palms.

[0096] Step S4: Perform live body recognition on the high-frequency palm image using a live body recognition model.

[0097] In this step, a live body judgment is performed on the extracted high-frequency palm image. The live body recognition model can use various types of models, such as deep learning models (convolutional neural network, CNN), traditional machine learning models (support vector machine, SVM; random forest, RF), etc.

[0098] For a deep learning model, such as using CNN, it can include multiple convolutional layers, pooling layers, and fully connected layers. The convolutional layer extracts features by sliding the convolutional kernel on the image, the pooling layer reduces the size of the feature map, and the fully connected layer maps the extracted features to the classification result. The model is trained with a large amount of palm image data (including live and non-live samples) to learn the feature patterns for distinguishing live and non-live.

[0099] For traditional machine learning models, the high-frequency palm image needs to be first converted into a feature vector, for example, using texture feature descriptors (such as local binary pattern, LBP; histogram of oriented gradients, HOG) or statistical features (such as mean, variance, etc.) as input.

[0100] Model training process:

[0101] Prepare a training data set, which contains a large number of palm image samples marked as "living" and "non-living". These samples are processed through the previous steps S1 to S3 to obtain high-frequency palm image samples.

[0102] The training data is divided into training set, validation set and test set, and allocated according to a certain ratio (such as 70:15:15).

[0103] For deep learning models, use the training set to train the model, select the appropriate loss function (such as cross entropy loss) and optimizer (such as Adam), use the validation set to adjust the model parameters and prevent overfitting during the training process, and finally evaluate the performance of the model on the test set.

[0104] For traditional machine learning models, training is performed using a training set, hyperparameters (such as kernel function parameters and penalty coefficients for SVM, number and depth of trees for random forest, etc.) are adjusted, the model is optimized based on the performance of the validation set, and evaluated on the test set.

[0105] In some embodiments, the liveness recognition model uses the palm image for supervision during training. In this embodiment, another HEAD is used for liveness judgment. Using the original palm image for supervision can ensure that the deviation in the model training process will not be too large, thereby ensuring the training effect.

[0106] In some embodiments, when training the liveness recognition model, different high-frequency parts are obtained by setting different high-frequency starting values ​​to improve the adaptability of the liveness recognition model. In the spectrum after Fourier transformation, the low-frequency part usually contains the general outline and overall shape information of the image, while the high-frequency part contains more details and texture information. The feature differences of different live and non-live palms in the high-frequency part may vary due to factors such as acquisition equipment, environmental conditions, and individual differences. By setting different high-frequency starting values, high-frequency parts of different ranges can be extracted, thereby covering more frequency bands that may contain effective distinguishing features, so that the model can learn richer feature representations during the training process, thereby improving the adaptability and robustness of the model. Determine multiple different high-frequency starting values. These starting values ​​can be selected based on experience or experiments. For example, circles with different radius ranges can be selected as the starting points of the high-frequency part. The features of the extracted different high-frequency parts are used as inputs and are input into the liveness recognition model separately or simultaneously for training. In this embodiment, more potential features can be mined during the training process by adjusting the starting value of the high-frequency part, thereby improving the adaptability and performance of the liveness recognition model, so that it can perform better in various complex practical application scenarios. In practice, the selection of high-frequency starting point values ​​and the architecture of the model can be continuously optimized according to specific experimental results and actual needs to achieve the best living body recognition effect.

[0107] Live body recognition process:

[0108] For the newly acquired palm image, after being processed through steps S1 to S3, the obtained high-frequency palm image is input into the trained live body recognition model.

[0109] Based on the learned features and patterns, the model outputs a classification result, usually a probability value (such as the live body probability) or a class label (live body or non-live body). According to this result, it is determined whether the palm is a live body. If it is a live body, the verification is passed; otherwise, the verification is rejected.

[0110] In this embodiment, starting from the acquisition of the palm image, the high-frequency information is extracted by using Fourier transform, and then converted back to the spatial domain to obtain a high-frequency palm image that emphasizes details. Finally, the trained live body recognition model is used for classification to achieve the live body detection of the palm, providing a biometric method that combines signal processing and machine learning, and providing a new and potential solution for the security authentication system.

[0111] Figure 4 It is a flowchart of the steps for extracting the high-frequency part in the embodiment of the present invention. As Figure 4 shown, the steps for extracting the high-frequency part in the embodiment of the present invention include:

[0112] Step S21: Perform two-dimensional Fourier transform on the palm image to obtain a complex matrix.

[0113] In this step, the two-dimensional Fourier transform is a mathematical tool for converting a two-dimensional image from the spatial domain to the frequency domain. For a two-dimensional discrete image f(x, y), where x and y represent the spatial coordinates of the image, the formula for its two-dimensional discrete Fourier transform (2DDFT) is:

[0114]

[0115] Here, M and N are the dimensions of the image, u and υ are the coordinates in the frequency domain, is the imaginary unit, is the complex exponential function.

[0116] Step S22: Calculate the spectrum information.

[0117] In this step, in order to visualize and analyze the information in the frequency domain, it is usually necessary to calculate the amplitude spectrum. For the complex matrix F(u, υ), the amplitude spectrum | F(u, υ) | is calculated by the formula:

[0118]

[0119] where Re(F(u, v)) is the real part of the complex number, and Im(F(u, v)) is the imaginary part of the complex number.

[0120] Step S23: Determine and isolate the high-frequency part from the spectral information according to a threshold value.

[0121] In this step, the determination of the threshold value is a key step, which can be selected through experiments or experience. A common method is to regard the center of the spectrum as the low-frequency part and the area far from the center as the high-frequency part. The threshold value can be determined according to the radius r of the frequency. For example, assuming that the center of the spectrum is (M / 2, N / 2), then for the frequency component (u, υ), the distance from it to the center

[0122] A distance threshold value r can be set, and the part where d > r is regarded as the high-frequency part. The size of r needs to be adjusted according to the characteristics of the palm image and the experimental results to ensure that the high-frequency part reflecting the living body characteristics can be effectively isolated.

[0123] When isolating the high-frequency part, the zero-frequency component (DC component) of the amplitude spectrum needs to be moved to the center to facilitate the calculation of the distance from the frequency component to the center. Then, the high-frequency part is isolated according to the set threshold value.

[0124] In this embodiment, the palm image is transformed into the frequency domain, the spectral information is calculated, and the high-frequency part is isolated according to the threshold value, providing key frequency-domain features for subsequent inverse transformation and living body recognition, and utilizing the characteristics of the Fourier transform to extract the information of the differences between living bodies and non-living bodies that may be contained in the palm image.

[0125] Figure 5 It is a flowchart of the steps for obtaining a high-frequency palm image in the embodiment of the present invention. As Figure 5 shown, the steps for obtaining a high-frequency palm image in the embodiment of the present invention include:

[0126] Step S31: Reconstruct a complex matrix according to the high-frequency part.

[0127] In this step, the high-frequency part obtained in step S2 is reconstructed into a complete complex matrix, and the low-frequency part (the part not extracted as the high-frequency part) can be set to zero or close to zero. This is because in the inverse Fourier transform, complete frequency-domain information is required, and only the high-frequency part is concerned, so the low-frequency part is weakened or eliminated to highlight the influence of the high-frequency part on the inverse transformation result.

[0128] Step S32: Apply the inverse Fourier transform algorithm to transform the high-frequency part in the frequency domain back to the spatial domain.

[0129] In this step, the inverse Fourier transform is applied to transform the complex matrix in the frequency domain back to the spatial domain. For the two-dimensional case, the two-dimensional discrete inverse Fourier transform (2DIDFT) or the two-dimensional fast Fourier inverse transform (2DIFFT) is used. The inverse transformation formula is:

[0130]

[0131] Among them, F(u, υ) is a complex matrix in the frequency domain, f(x, y) is an image matrix in the spatial domain, M and N are the dimensions of the image, is the imaginary unit.

[0132] Step S33: Process the result in the spatial domain to obtain a high-frequency palm image.

[0133] In this step, the result obtained by the inverse Fourier transform is a complex matrix, while what is needed is a displayable image. Therefore, the complex matrix needs to be converted into a real matrix. Usually, the real part or amplitude of the complex matrix can be taken as the final image pixel value because the real part or amplitude of the complex number represents the intensity of the signal.

[0134] In this embodiment, the high-frequency part spectral information extracted from step S2 finally obtains a high-frequency palm image that can be used for subsequent live body recognition after reconstructing the complex matrix, performing inverse Fourier transform, and processing the result. This image emphasizes the high-frequency detail information in the original palm image and provides more discriminative image data for the subsequent live body recognition steps.

[0135] Figure 6 is a flowchart of the steps for performing live body recognition on a high-frequency palm image in an embodiment of the present invention. As Figure 6 shown, the steps for performing live body recognition on a high-frequency palm image in an embodiment of the present invention include:

[0136] Step S41: Extract features from the high-frequency palm image to obtain palm features.

[0137] In this step, feature extraction is to convert the image into a set of vectors that can represent the features of the image, so that the subsequent model can better process and classify it. For high-frequency palm images, various feature extraction methods can be used. These methods aim to capture information such as texture, shape, and structure in the image, so as to find the differences between live and non-live palms.

[0138] Commonly used feature extraction methods include:

[0139] Local Binary Patterns (LBP): It is a texture descriptor. By comparing the gray values of the central pixel with its neighboring pixels, the result is encoded as a binary number, and then the frequencies of different patterns are counted. It has strong descriptive ability for texture features and can effectively capture texture details in high-frequency palm images.

[0140] Histogram of Oriented Gradients (HOG): Calculate the histogram of gradient directions in local regions of an image. By dividing the image into small connected regions (cell units), the gradient direction and intensity of each unit are statistically calculated to form a feature vector. For high-frequency palm images containing edge information, HOG can extract the direction and intensity information of the edges.

[0141] Scale-Invariant Feature Transform (SIFT): Extract key points and their descriptors in an image. These key points are scale and rotation invariant and are very effective for detecting key feature points in the palm (such as parts with drastic texture changes). Although the computational cost is relatively high, it may be useful in some scenarios with high precision requirements.

[0142] Convolutional Neural Networks (CNN) feature extraction: If a deep learning method is used, the high-frequency palm image can be input into a pre-trained CNN, and its convolutional layers are used to automatically extract features. Different convolutional layers extract different levels of features. Shallow layers usually extract low-level features such as edges and textures, while deep layers extract more abstract semantic features.

[0143] Step S42: Input the palm features into the liveness recognition model.

[0144] In this step, the liveness recognition model can be a classifier based on traditional machine learning algorithms or deep learning algorithms. Traditional machine learning algorithms include Support Vector Machines (SVM), Random Forest, K-Nearest Neighbors (KNN), etc. They take the feature vector as input and classify it according to the learned decision boundary or rules. Deep learning algorithms use neural network structures and automatically learn the complex relationships between features through multiple hidden layers.

[0145] For traditional machine learning algorithms, the model needs to be trained first, using labeled training data (including feature vectors of live and non-live palms) to learn the classification rules. For deep learning algorithms, they are usually pre-trained using large-scale data and then the fine-tuning technique can be used to apply them to the palm liveness recognition task.

[0146] Step S43: Obtain the liveness recognition result.

[0147] In this step, when the palm features are input into the trained liveness recognition model, the model will output a classification result according to the rules or feature relationships it has learned. For a binary classification problem, the result is usually a class label (such as live or non-live) or a probability value indicating the probability of being live.

[0148] In this embodiment, features are extracted from high-frequency palm images, and these features are input into the liveness recognition model to finally obtain the liveness recognition result of the palm. In practical applications, a large amount of training data is required to fully train the model to ensure its accuracy and reliability. At the same time, different feature extraction methods and classification models can be selected according to specific requirements to achieve the best liveness recognition effect.

[0149] Figure 7 It is a schematic structural diagram of a high-frequency-based palm liveness detection system in an embodiment of the present invention. As Figure 7 shown, a high-frequency-based palm liveness detection system in an embodiment of the present invention includes:

[0150] An acquisition module for acquiring a first image set and a second image set; both the first image set and the second image set include RGB images and speckle infrared images; the first image set and the second image set are obtained at different times;

[0151] A recognition module for inputting the first image set and the second image set into a time series model, and identifying the range of the transparent object and the boundary of the transparent object according to the differences between the first image set and the second image set;

[0152] A correction module for correcting the depth values within the range of the transparent object according to the depth value of the boundary of the transparent object to obtain the final depth map of the transparent object.

[0153] Specifically, the acquisition module undertakes the task of image acquisition and is specifically responsible for capturing two sets of image collections at different time points, namely the first image set and the second image set. Each set of image collections covers both RGB images and speckle infrared images at the same time. The RGB images capture rich color and texture information of the scene, while the speckle infrared images utilize the characteristics of the infrared band to reflect the thermal radiation and speckle distribution of the object, providing a multi-dimensional data basis for subsequent transparent object recognition and depth analysis.

[0154] As a key analysis unit, the recognition module receives the two sets of image collections transmitted by the acquisition module and inputs them into the time series model. The time series model has the ability to process time series data. By deeply exploring the subtle differences generated by the two sets of images over time, such as the changes in the optical characteristics of the object over time, it accurately locates the range of the transparent object in the scene and outlines its clear boundary with the surrounding environment, defining the operation area for depth correction.

[0155] The calibration module focuses on the optimization of depth information processing. Taking the depth values of the boundaries of transparent objects output by the recognition module as an important reference, considering that the internal depth measurement of transparent objects often has large errors, this module uses this relatively reliable boundary depth data and applies specific algorithms to adjust and calibrate the depth values within the range of transparent objects, and finally outputs an accurate and error-free final depth map of transparent objects, providing accurate depth data support for subsequent applications such as robot grasping and scene three-dimensional reconstruction.

[0156] In this embodiment, by obtaining a palm image, performing Fourier transform on it to obtain spectral information and extracting the high-frequency part, then performing inverse Fourier transform to obtain a high-frequency palm image, and finally using a living body recognition model to complete the living body recognition of the palm based on this high-frequency palm image, it has the advantages of general equipment, anti-interference, effective feature extraction, strong learning ability, scalability, high security and good reliability.

[0157] In an embodiment of the present invention, there is also provided a high-frequency-based palm living body detection device, including a processor and a memory in which executable instructions of the processor are stored. Wherein, the processor is configured to execute the steps of a high-frequency-based palm living body detection method by executing the executable instructions.

[0158] As above, in this embodiment, by obtaining a palm image, performing Fourier transform on it to obtain spectral information and extracting the high-frequency part, then performing inverse Fourier transform to obtain a high-frequency palm image, and finally using a living body recognition model to complete the living body recognition of the palm based on this high-frequency palm image, it has the advantages of general equipment, anti-interference, effective feature extraction, strong learning ability, scalability, high security and good reliability.

[0159] Those skilled in the art to which the present invention pertains can understand that various aspects of the present invention can be implemented as a system, a method, or a program product. Therefore, various aspects of the present invention can be specifically implemented in the following forms, namely: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or an implementation combining hardware and software aspects, which can be collectively referred to as "circuit", "module", or "platform" here.

[0160] Figure 8 It is a schematic structural diagram of a high-frequency-based palm living body detection device in an embodiment of the present invention. The following will refer to Figure 8 to describe the electronic device 600 according to this embodiment of the present invention. Figure 8 The electronic device 600 shown is only an example and should not impose any limitations on the functions and usage scope of the embodiments of the present invention.

[0161] As Figure 8As shown, the electronic device 600 is presented in the form of a general-purpose computing device. The components of the electronic device 600 may include, but are not limited to: at least one processing unit 610, at least one storage unit 620, a bus 630 connecting different platform components (including the storage unit 620 and the processing unit 610), a display unit 640, etc.

[0162] Among them, the storage unit stores program code, and the program code can be executed by the processing unit 610, so that the processing unit 610 executes the steps according to various exemplary embodiments of the present invention described in the above-mentioned part of the high-frequency-based palm liveness detection method in this specification. For example, the processing unit 610 can execute as Figure 1 the steps shown in.

[0163] The storage unit 620 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 6201 and / or a cache storage unit 6202, and may further include a read-only storage unit (ROM) 6203.

[0164] The storage unit 620 may also include a program / utility 6204 having a set (at least one) of program modules 6205. Such program modules 6205 include, but are not limited to: an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include the implementation of a grid environment.

[0165] The bus 630 may represent one or more of several types of bus structures, including a storage unit bus or a storage unit controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any bus structure in a variety of bus structures.

[0166] The electronic device 600 can also communicate with one or more external devices 700 (such as a keyboard, a pointing device, a Bluetooth device, etc.), can also communicate with one or more devices that enable a user to interact with the electronic device 600, and / or communicate with any device that enables the electronic device 600 to communicate with one or more other computing devices (such as a router, a modem, etc.). Such communication can be carried out through an input / output (I / O) interface 650. And, the electronic device 600 can also communicate with one or more grids (such as a local area network (LAN), a wide area network (WAN), and / or a public grid, such as the Internet) through a grid adapter 660. The grid adapter 660 can communicate with other modules of the electronic device 600 through the bus 630. It should be understood that although Figure 8which is not shown in the figure. Other hardware and / or software modules may be used in combination with the electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage platforms, etc.

[0167] In an embodiment of the present invention, there is also provided a computer-readable storage medium for storing a program, and when the program is executed, it implements the steps of a high-frequency-based palm liveness detection method. In some possible implementation manners, various aspects of the present invention may also be implemented in the form of a program product, which includes program code. When the program product runs on a terminal device, the program code is used to cause the terminal device to execute the steps according to various exemplary embodiments of the present invention described in the above-mentioned part of the high-frequency-based palm liveness detection method in this specification.

[0168] As shown above, in this embodiment, by acquiring a palm image, performing Fourier transform on it to obtain spectral information and extracting the high-frequency part, then performing inverse Fourier transform to obtain a high-frequency palm image, and finally using a liveness recognition model to complete the liveness recognition of the palm based on this high-frequency palm image, it has the advantages of device generality, anti-interference, effective feature extraction, strong learning ability, scalability, high security, and good reliability.

[0169] Figure 9 is a schematic structural diagram of the computer-readable storage medium in an embodiment of the present invention. Refer to Figure 9 As shown, a program product 800 for implementing the above method according to an embodiment of the present invention is described. It may be a portable compact disc read-only memory (CD-ROM) and includes program code, and can run on a terminal device, such as a personal computer. However, the program product of the present invention is not limited thereto. In this document, the readable storage medium may be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, apparatus, or device.

[0170] The program product may adopt any combination of one or more readable media. The readable media may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0171] A computer-readable storage medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries readable program code. Such a propagated data signal may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. The readable storage medium may also be any readable medium other than the readable storage medium, which can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0172] The program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages such as Java, C++, etc., and also including conventional procedural programming languages such as the "C" language or similar programming languages. The program code may be executed entirely on the user's computing device, partially on the user's device, executed as a stand-alone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user's computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., by using an Internet service provider to connect through the Internet).

[0173] In this embodiment, by obtaining a palm image, performing Fourier transform on it to obtain spectral information and extracting the high-frequency part, then obtaining a high-frequency palm image through inverse Fourier transform, and finally using a living body recognition model to complete the living body recognition of the palm based on the high-frequency palm image, it has the advantages of device generality, anti-interference, effective feature extraction, strong learning ability, scalability, high security, and good reliability.

[0174] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The same or similar parts among the various embodiments may be referred to each other. The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0175] The specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various deformations or modifications within the scope of the claims, which do not affect the essence of the present invention.

Claims

1. A palm liveness detection method based on high frequency, characterized in that: include: Step S1: Acquire a palm image; Step S2: Perform Fourier transform on the palm image to obtain spectrum information and extract the high-frequency part; Step S3: performing inverse Fourier transform on the high-frequency part to obtain a high-frequency palm image; Step S4: Performing liveness recognition on the high-frequency palm image using a liveness recognition model.

2. The high-frequency palm liveness detection method according to claim 1, characterized in that: The palm image obtained in step S1 is an infrared image.

3. The high-frequency palm liveness detection method according to claim 1, characterized in that: Step S2 includes: Step S21: performing a two-dimensional Fourier transform on the palm image to obtain a complex matrix; Step S22: Calculate spectrum information; Step S23: determining and separating the high frequency part from the spectrum information according to the threshold.

4. The high-frequency palm liveness detection method according to claim 1, characterized in that: Step S3 includes: Step S31: reconstructing a complex matrix according to the high frequency part; Step S32: Applying the inverse Fourier transform algorithm to convert the high frequency part of the frequency domain back to the spatial domain; Step S33: Process the result in the spatial domain to obtain a high-frequency palm image.

5. The high-frequency palm liveness detection method according to claim 1, characterized in that: Step S4 includes: Step S41: extracting features from the high-frequency palm image to obtain palm features; Step S42: inputting the palm features into a living body recognition model; Step S43: Obtaining the liveness recognition result.

6. The high-frequency-based palm liveness detection method according to claim 1, characterized in that: The liveness recognition model uses the palm image as supervision during training.

7. The high-frequency palm liveness detection method according to claim 1, characterized in that: When training the liveness recognition model, different high-frequency starting point values ​​are set to obtain different high-frequency parts, thereby improving the adaptability of the liveness recognition model.

8. A high-frequency-based palm liveness detection system, used to implement the high-frequency-based palm liveness detection method according to any one of claims 1 to 7, characterized in that: include: An acquisition module, used for acquiring a palm image; An extraction module, used for performing Fourier transform on the palm image to obtain spectrum information and extract the high-frequency part; An inverse transform module, used for performing inverse Fourier transform on the high-frequency part to obtain a high-frequency palm image; The recognition module is used to perform liveness recognition on the high-frequency palm image using a liveness recognition model.

9. A high-frequency palm liveness detection device, characterized in that: include: processor; a memory storing executable instructions of the processor; The processor is configured to execute the steps of the high-frequency-based palm liveness detection method according to any one of claims 1 to 7 by executing the executable instructions.

10. A computer-readable storage medium for storing a program, characterized in that: When the program is executed, the steps of the high-frequency-based palm liveness detection method described in any one of claims 1 to 7 are implemented.

Images