Liquid crystal display fingerprint module reading method, system and device and storage medium

Abstract

The invention discloses a liquid crystal display screen fingerprint module reading method, system and device and a storage medium. The method comprises the following steps: acquiring an initial fingerprint image of a liquid crystal display screen fingerprint module; performing preprocessing operation on the initial fingerprint image to obtain a second fingerprint image; inputting the second fingerprint image into a pre-trained liquid crystal display screen fingerprint reading processing model to obtain a third fingerprint image; inputting the initial fingerprint image and the third fingerprint image into a pre-trained fingerprint image correction model based on high dynamic range image fusion when the effective feature points contained in the third fingerprint image do not meet a preset feature point threshold value, and obtaining a final result image. And performing fingerprint reading according to the final result image. The method has the following effects: when fingerprint damage or environment illumination change causes an image quality problem, the detail features of the fingerprint can be accurately extracted and matched.

Description

Technical Field

[0001] The present invention relates to the technical field of liquid crystal display screens, and in particular to a liquid crystal display screen fingerprint module reading method, system, device and storage medium. Background Art

[0002] At present, LCD fingerprint modules, as a highly integrated biometric technology device, have demonstrated their irreplaceable and important role in the field of personal identity verification and security authentication. These modules build a solid security line for various devices and systems by accurately reading and verifying individual fingerprint information, significantly improving the overall security performance.

[0003] In one prior art, LCD fingerprint modules use some basic image enhancement technologies in image processing, such as contrast stretching and histogram equalization. Contrast stretching technology enhances the contrast of fingerprint images by adjusting the distribution range of pixel values ​​in the image, making the boundaries between fingerprint ridges and valleys clearer, while histogram equalization technology redistributes the brightness distribution of the image to make the overall brightness of the image more uniform, reducing image quality problems caused by uneven lighting.

[0004] Although the existing technology has made certain progress in image processing, it is still unable to accurately extract and match the detailed features of fingerprints when faced with some complex scenarios, such as when fingerprint damage or changes in ambient lighting cause image quality problems. Summary of the invention

[0005] The present invention provides a liquid crystal display fingerprint module reading method, system, device and storage medium to solve the problem that the prior art cannot accurately extract and match the detailed features of the fingerprint when facing some complex scenes, such as when the fingerprint is damaged or the ambient light changes cause image quality problems.

[0006] In the first aspect, in order to solve the above technical problems, the present invention provides a method for reading a fingerprint module on a liquid crystal display screen, comprising:

[0007] Get the initial fingerprint image of the LCD fingerprint module;

[0008] Performing a preprocessing operation on the initial fingerprint image to obtain a second fingerprint image;

[0009] Inputting the second fingerprint image into a pre-trained LCD display fingerprint reading processing model to obtain a third fingerprint image;

[0010] When it is determined that the valid feature points contained in the third fingerprint image do not meet the preset feature point threshold, the initial fingerprint image and the third fingerprint image are input into a pre-trained fingerprint image correction model based on high dynamic range image fusion to obtain a final result image, and fingerprint reading is performed based on the final result image.

[0011] In an optional implementation, the training process of the LCD display fingerprint reading processing model includes:

[0012] Building a convolutional neural network model based on the grayscale, texture and edge features of the historical fingerprint image, and training the convolutional neural network model;

[0013] When the training times of the convolutional neural network model is greater than or equal to the preset maximum training times, or when the loss function value of the convolutional neural network model is less than the preset loss threshold, the training is determined to be completed and a liquid crystal display fingerprint reading processing model is obtained.

[0014] In an optional implementation, the training process of the fingerprint image correction model based on high dynamic range image fusion includes:

[0015] Building a generative adversarial network model based on the features of high dynamic range images and a multi-image fusion algorithm, and training the generative adversarial network model;

[0016] When the training times of the generative adversarial network model is greater than or equal to the preset maximum training times, or when the loss function value of the generative adversarial network model is less than the preset loss threshold, the training is determined to be completed, and a fingerprint image correction model based on high dynamic range image fusion is obtained.

[0017] In an optional implementation, the preprocessing operation on the initial fingerprint image includes:

[0018] Noise reduction is achieved through the following linear filtering formula:

[0019]

[0020] Where N is the total number of pixels in the filter window. For a (2a+1)×(2b+1) window, N=(2a+1)(2b+1), a and b are the radii of the filter window in the horizontal and vertical directions, respectively. m and n are the coordinate offsets relative to the center pixel (x, y). I′(x, y) is the pixel value at the coordinate (x, y) of the image after linear filtering and denoising.

[0021] Perform Fourier transform on the denoised image:

[0022]

[0023] Among them, u and v are frequency variables, u represents the frequency of the image in the horizontal direction, v represents the frequency of the image in the vertical direction, j is the imaginary unit, It means Fourier transform of I′(x, y);

[0024] Filter out high-frequency components of the Fourier transformed image:

[0025] F′(u,v)=F(u,v)·L(u,v)

[0026] Among them, L(u, v) is a low-pass filter, · represents the dot product operation, Where D is the cutoff frequency, Represents the distance from the origin of the frequency domain to the point (u, v);

[0027] Perform inverse Fourier transform on the frequency domain image after filtering out the high-frequency components, and convert the image from the frequency domain back to the spatial domain:

[0028]

[0029] Where I″(x, y) is the pixel value of the denoised and smoothed image at coordinate (x, y). is the inverse Fourier transform.

[0030] In an optional implementation, the step of inputting the second fingerprint image into a pre-trained LCD display fingerprint reading processing model to obtain a third fingerprint image includes:

[0031] Performing image segmentation and threshold processing on the second fingerprint image to obtain an enhanced fingerprint image;

[0032] The enhanced fingerprint image is thinned and integrated to obtain the third fingerprint image.

[0033] In an optional embodiment, the method comprises:

[0034] Feature point detection is performed using the following formula:

[0035]

[0036] Where S(x, y) represents the intensity of the feature point at the image coordinate (x, y), I x and I y are the gradient values ​​of the image in the x and y directions respectively, and θ is the preset feature point threshold;

[0037] When S(x, y)>θ is satisfied, the point (x, y) is identified as a feature point;

[0038] When the counted number of feature points is less than a preset feature point number threshold, it is determined that the valid feature points included in the third fingerprint image do not meet the requirements.

[0039] In an optional implementation, the inputting the initial fingerprint image and the third fingerprint image into a pre-trained fingerprint image correction model based on high dynamic range image fusion to obtain a final result image includes:

[0040] Extracting features from the initial fingerprint image and the third fingerprint image respectively to obtain a feature set including image edge intensity features and texture features;

[0041] Based on the first weight coefficient and the second weight coefficient, the feature set is fused to obtain a fused edge feature and a fused texture feature;

[0042] Based on the third weight coefficient and the fourth weight coefficient, image reconstruction is performed on the fused edge features and the fused texture features to obtain a final result image.

[0043] In a second aspect, the present invention provides a liquid crystal display fingerprint module reading system, comprising:

[0044] A data acquisition module is used to acquire an initial fingerprint image of a fingerprint module on a liquid crystal display;

[0045] A data processing module, used for performing a preprocessing operation on the initial fingerprint image to obtain a second fingerprint image;

[0046] A fingerprint reading processing module, used for inputting the second fingerprint image into a pre-trained liquid crystal display fingerprint reading processing model to obtain a third fingerprint image;

[0047] The fingerprint image optimization module is used to input the initial fingerprint image and the third fingerprint image into a pre-trained fingerprint image correction model based on high dynamic range image fusion to obtain a final result image when it is determined that the valid feature points contained in the third fingerprint image do not meet a preset feature point threshold, and perform fingerprint reading based on the final result image.

[0048] In a third aspect, the present invention further provides an electronic device, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the liquid crystal display fingerprint module reading method described in any one of the above is implemented.

[0049] In a fourth aspect, the present invention further provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned LCD display fingerprint module reading methods.

[0050] Compared with the prior art, the present invention has the following beneficial effects: The present invention proposes an innovative reading method for the image quality problems and feature extraction challenges faced by the fingerprint module of the liquid crystal display in complex scenarios (such as fingerprint damage or changes in ambient light). The method first captures the initial fingerprint image and performs image preprocessing operations on it to improve image quality, reduce noise and enhance contrast. Subsequently, the preprocessed image is further processed using a pre-trained deep learning model to extract the key feature points of the fingerprint. If the model determines that the extracted feature points do not meet the preset threshold requirements, the fingerprint image correction model based on high dynamic range image fusion is automatically triggered, and the initial image is intelligently fused with the preliminarily processed image to obtain a final result image containing more abundant effective feature points. This process improves the accuracy of fingerprint recognition and the robustness of the system under different environmental conditions, providing users with a more stable and efficient fingerprint reading experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a flow chart of a method for reading a fingerprint module on a liquid crystal display screen provided by the first embodiment of the present invention;

[0052] Figure 2 It is a schematic diagram of the structure of a liquid crystal display fingerprint module reading system provided by the second embodiment of the present invention. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] Reference Figure 1 The first embodiment of the present invention provides a method for reading a fingerprint module on a liquid crystal display screen, comprising the following steps:

[0055] S11, obtaining an initial fingerprint image of the LCD fingerprint module;

[0056] S12, performing a preprocessing operation on the initial fingerprint image to obtain a second fingerprint image;

[0057] S13, inputting the second fingerprint image into a pre-trained liquid crystal display fingerprint reading processing model to obtain a third fingerprint image;

[0058] S14, when it is determined that the valid feature points contained in the third fingerprint image do not meet the preset feature point threshold, the initial fingerprint image and the third fingerprint image are input into a pre-trained fingerprint image correction model based on high dynamic range image fusion to obtain a final result image, and fingerprint reading is performed based on the final result image.

[0059] In step S11, obtaining the initial fingerprint image of the LCD fingerprint module is the key starting point of the entire identification process. The initial fingerprint image is the original digital representation of the fingerprint texture captured by the sensor built into the fingerprint module when the user's finger directly touches the LCD screen. This process is achieved through high-precision optical or capacitive sensors, which can sense the tiny bumps on the fingerprint surface and convert them into processable image data.

[0060] In order to obtain a high-quality initial fingerprint image, the system first ensures that the lighting, temperature, humidity and other conditions of the collection environment are within an appropriate range to reduce the interference of external factors on the image quality. At the same time, the user needs to place the finger correctly in the fingerprint collection area according to the instructions, maintain appropriate pressure and stable posture to ensure the integrity and clarity of the fingerprint image.

[0061] Once the initial fingerprint image is successfully captured, it will be immediately used in the subsequent image preprocessing and feature extraction steps. In these steps, the initial fingerprint image will undergo a series of algorithmic processing, such as noise suppression, contrast enhancement, image sharpening, etc., to improve image quality and highlight fingerprint features. In addition, the initial fingerprint image can also be used as a reference to compare or fuse with the image processed by the deep learning model to further improve the accuracy and robustness of fingerprint recognition.

[0062] In step S12, a series of preprocessing operations are performed on the initial fingerprint image to improve the image quality and lay a solid foundation for subsequent feature extraction and matching. This process first achieves noise reduction through a linear filtering formula, which is based on the average pixel value calculation within the filter window. By adjusting the window radius in the horizontal and vertical directions (i.e., and), the pixel values ​​around the center pixel are weighted averaged to suppress random noise in the image and improve the image clarity and signal-to-noise ratio. The denoised image is smoother, reducing the problem of fingerprint feature blurring caused by noise interference.

[0063] Subsequently, the preprocessing process uses Fourier transform to convert the denoised image from the spatial domain to the frequency domain. Fourier transform is a powerful mathematical tool that can decompose image signals into components of different frequencies, so that high-frequency (details, edges) and low-frequency (background, smooth areas) information in the image can be separated. In the frequency domain, a low-pass filter is designed and applied to the dot multiplication operation to filter out high-frequency components. This process essentially removes small noise and unnecessary details in the image, further enhancing the contrast between the fingerprint ridges and valleys, making the fingerprint lines clearer and more discernible.

[0064] Finally, the frequency domain image after filtering out the high-frequency components is subjected to inverse Fourier transform, and converted back to the spatial domain to obtain the processed second fingerprint image. This process not only completes the spatial reconstruction of the image, but also achieves the smoothing and denoising effects of the image, making the second fingerprint image more suitable for subsequent fingerprint feature extraction and analysis. Through this series of preprocessing steps, not only the overall quality of the fingerprint image is improved, but also the accuracy and robustness of the fingerprint recognition system are significantly improved.

[0065] In one implementation, the preprocessing operation on the initial fingerprint image includes:

[0066] Noise reduction is achieved through the following linear filtering formula:

[0067]

[0068] Where N is the total number of pixels in the filter window. For a (2a+1)×(2b+1) window, N=(2a+1)(2b+1), a and b are the radii of the filter window in the horizontal and vertical directions, respectively. m and n are the coordinate offsets relative to the center pixel (x, y). I′(x, y) is the pixel value at the coordinate (x, y) of the image after linear filtering and denoising.

[0069] Perform Fourier transform on the denoised image:

[0070]

[0071] Among them, u and v are frequency variables, u represents the frequency of the image in the horizontal direction, v represents the frequency of the image in the vertical direction, j is the imaginary unit, It means Fourier transform of I′(x, y);

[0072] Filter out high-frequency components of the Fourier transformed image:

[0073] F′(u,v)=F(u,v)·L(u,v)

[0074] Among them, L(u, v) is a low-pass filter, · represents the dot product operation, Where D is the cutoff frequency, Represents the distance from the origin of the frequency domain to the point (u, v);

[0075] Perform inverse Fourier transform on the frequency domain image after filtering out the high-frequency components, and convert the image from the frequency domain back to the spatial domain:

[0076]

[0077] Where I″(x, y) is the pixel value of the denoised and smoothed image at coordinate (x, y). is the inverse Fourier transform.

[0078] It should be noted that the above preprocessing operation is a crucial step in the LCD display fingerprint module reading method, which is directly related to the accuracy and reliability of subsequent fingerprint feature extraction. The following is a detailed explanation of each step of the preprocessing process:

[0079] First, the initial fingerprint image is denoised using a linear filtering formula. The purpose of this step is to reduce random noise in the image and improve image clarity and signal-to-noise ratio. Linear filtering is based on the average pixel value calculation within the filter window. By adjusting the window radius in the horizontal and vertical directions, the pixel values ​​around the center pixel are weighted averaged. Specifically, for each pixel within the filter window, its weighted contribution is calculated based on its coordinate offset (x and y) from the center pixel, and the sum is used to obtain the new value of the center pixel. This method can effectively suppress high-frequency noise in the image while retaining the texture information of the fingerprint.

[0080] The denoised image is then Fourier transformed to convert the image from the spatial domain to the frequency domain. Fourier transform is a mathematical transformation method that decomposes the image into components of different frequencies to facilitate subsequent frequency domain processing of the image. Through Fourier transform, low-frequency components (such as fingerprint texture) and high-frequency components (such as noise and details) in the image can be separated.

[0081] In the frequency domain, high-frequency components usually correspond to noise and details in the image, while low-frequency components contain the main information of the image. In order to further remove noise and smooth the image, it is necessary to filter out the high-frequency components of the Fourier transformed image. This step is achieved by applying a low-pass filter, which allows low-frequency components to pass through and suppresses high-frequency components.

[0082] Finally, the frequency domain image after filtering out the high-frequency components is subjected to inverse Fourier transform to convert the image from the frequency domain back to the spatial domain. Inverse Fourier transform is the inverse process of Fourier transform, which uses the transformed frequency domain information to reconstruct the original image. Through inverse Fourier transform, the processed second fingerprint image is obtained, which effectively removes noise while retaining the fingerprint texture information, providing a high-quality image foundation for subsequent feature extraction and recognition.

[0083] In summary, by performing a series of preprocessing operations on the initial fingerprint image, such as linear filtering denoising, Fourier transform, filtering out high-frequency components and inverse Fourier transform, the quality of the fingerprint image can be significantly improved, laying a solid foundation for subsequent feature extraction and recognition.

[0084] In step S13, the third fingerprint image is the result of deep processing and feature enhancement of the second fingerprint image, which is intended to further improve the quality of the fingerprint image so as to more accurately extract and identify fingerprint features. This step is achieved by inputting the second fingerprint image into a pre-trained LCD display fingerprint reading processing model.

[0085] The LCD display fingerprint reading processing model is built based on a deep learning algorithm, which can optimize the unique optical characteristics of LCD displays and environmental interference (such as screen reflection, uneven fingerprint pressing, etc.). Through a complex network structure and a large amount of training data learning, the model can automatically identify and compensate for these interference factors, thereby outputting a higher quality third fingerprint image. This process not only further reduces image noise, but also significantly enhances the contrast between fingerprint ridges and valleys, making fingerprint features clearer and more discernible.

[0086] In one embodiment, the training process of the LCD display fingerprint reading processing model includes:

[0087] Building a convolutional neural network model based on the grayscale, texture and edge features of the historical fingerprint image, and training the convolutional neural network model;

[0088] When the training times of the convolutional neural network model is greater than or equal to the preset maximum training times, or when the loss function value of the convolutional neural network model is less than the preset loss threshold, the training is determined to be completed and a liquid crystal display fingerprint reading processing model is obtained.

[0089] It should be noted that this process involves some key concepts and technologies in the field of machine learning and deep learning. The following is a detailed explanation of the process:

[0090] Before training a convolutional neural network (CNN), the historical fingerprint images need to be preprocessed. This includes converting the images to grayscale images, as grayscale images can reduce computational complexity while retaining enough information for feature extraction. In addition, steps such as image scaling, normalization, and enhancement are also included to improve the generalization and robustness of the model. In addition, one of the core functions of a convolutional neural network is to automatically extract features from input data. In the case of fingerprint images, grayscale, texture, and edge features are key to identifying fingerprints. CNN achieves this through a series of convolutional layers, activation functions, and pooling layers. Convolutional layers are used to extract local features, activation functions (such as ReLU) are used to introduce nonlinearity, and pooling layers are used to reduce the spatial dimension of features.

[0091] The training process involves adjusting the weights of the CNN to minimize the loss function. The loss function is a measure of the difference between the model's prediction and the true label. In fingerprint recognition, commonly used loss functions include cross entropy loss and mean square error loss. The training process uses the backpropagation algorithm to update the weights by calculating the gradient of the loss function with respect to the weights. At the same time, in order to avoid overfitting, an early stopping mechanism is set. When the number of training times (i.e., the number of iterations) of the model reaches the preset maximum number of training times, or the model's loss function value drops below the preset loss threshold, the training process will terminate early. This ensures that the model will not continue to train after reaching a certain performance, thereby avoiding overfitting on the training data.

[0092] After training, the model needs to be evaluated to ensure that it has good generalization ability. This involves evaluating the performance of the model on an independent validation set or test set. Common evaluation metrics include accuracy, recall, precision, and F1 score.

[0093] After obtaining the preliminary model, some optimization steps are required to further improve the performance of the model, including adjusting the network structure, hyperparameter tuning, and data augmentation.

[0094] Finally, the trained model can be deployed to the actual LCD display fingerprint reading device. In practical applications, the model needs to be able to quickly and accurately process the user's fingerprint image and provide real-time recognition results.

[0095] In one implementation, the step of inputting the second fingerprint image into a pre-trained liquid crystal display fingerprint reading processing model to obtain a third fingerprint image includes:

[0096] Performing image segmentation and threshold processing on the second fingerprint image to obtain an enhanced fingerprint image;

[0097] The enhanced fingerprint image is thinned and integrated to obtain the third fingerprint image.

[0098] It should be noted that the second fingerprint image B i，j Divide into an N×N uniform grid:

[0099] B′ i，j = {p∈B i,j |I2(p)>T j}

[0100] Where p represents the pixel position, T j is the threshold set for the jth block, B′ i，j Represents the second fingerprint image B after preprocessing i，j After block processing, a specific block B i，j Exceeding the threshold value T j A collection of pixels;

[0101] For grid block B′ i，j To perform grayscale stretching, the formula is:

[0102]

[0103] Where min(B′ i，j ) and max(B′ i，j ) are blocks B′ i，j The minimum and maximum grayscale values ​​in , L is the number of gray levels (e.g., for an 8-bit image, L=256), I′ i，j (x, y) represents the image B' i，j The image block obtained after grayscale stretching;

[0104] All processed blocks I′ i，j Integrate into a complete image I region :

[0105]

[0106] Among them, I region Represents the entire image area, which contains the information of all processed image blocks;

[0107] to I region Image weight coefficient calculation:

[0108]

[0109] where each pixel is assigned a weight coefficient w1 and w2 are preset weight values;

[0110] Perform weighted averaging on the image to enhance features:

[0111]

[0112] Finally, the final visualized fingerprint image is obtained through the following steps:

[0113] The enhanced image is convolved with a Gabor filter to extract fingerprint features. The convolution result is binarized to obtain a refined image to clearly display the fingerprint ridges. The refined image is histogram equalized to obtain a visualized fingerprint image.

[0114] In step S14, it is crucial to ensure that the fingerprint image contains a sufficient number of valid feature points, because these feature points are the cornerstone of subsequent fingerprint comparison and recognition. When the third fingerprint image (the image after deep processing and feature enhancement) is judged to contain a number of valid feature points that does not meet the preset feature point threshold, it means that the image may fail to fully display the detailed information of the fingerprint due to various reasons (such as poor fingerprint quality, severe environmental interference, etc.), thereby affecting the accuracy of recognition. At this time, introducing a fingerprint image correction model based on high dynamic range (HDR) image fusion becomes an effective solution.

[0115] The determination process scans the third fingerprint image through a specific feature point detection algorithm (such as a gradient-based feature point detection method), which uses the gradient values ​​of the image in the horizontal and vertical directions to calculate the feature point intensity at each pixel. By comparing the calculated feature point intensity with the preset feature point threshold, when the feature point intensity is greater than or equal to the threshold, the point is considered to be a valid feature point. Subsequently, the number of valid feature points in the entire image is counted and compared with the preset feature point number threshold. If the number of feature points obtained by counting is less than the threshold, it is determined that the valid feature points contained in the third fingerprint image do not meet the requirements and need to be further input into the fingerprint image correction model based on high dynamic range image fusion for processing.

[0116] The training process of the fingerprint image correction model based on high dynamic range image fusion is complex and sophisticated. The core of the training process is to build a generative adversarial network (GAN) model that can process HDR image features and realize multi-image fusion. The GAN model is well-known for its unique adversarial training mechanism. Through the continuous game between the generator and the discriminator, the quality of the generated image is gradually optimized. In this context, the characteristics of HDR images are combined with the multi-image fusion algorithm to generate high-quality fingerprint images that retain global illumination details and enhance local texture features.

[0117] First, according to the characteristics of HDR images and the requirements of fingerprint image correction, the network structures of the generator and discriminator are designed. The generator usually contains multiple convolutional layers, upsampling layers, and possible residual connections to capture the details in the image and generate realistic fingerprint images. The discriminator is responsible for distinguishing the generated fingerprint image from the real fingerprint image, prompting the generator to continuously improve the generation quality.

[0118] Then the fusion algorithm is integrated. In the generator, a multi-image fusion algorithm is integrated, so that the model can automatically learn how to effectively fuse the dominant information in different fingerprint images (such as the global illumination information of the initial fingerprint image and the local detail information of the third fingerprint image) during the training process. This involves multiple steps of feature extraction, feature matching, and feature fusion.

[0119] During the training process, the convergence of the model is determined by monitoring the number of training times and the loss function value. When the number of training times reaches the preset maximum number of training times, or the loss function value drops below the preset loss threshold, the training is considered complete, and the final fingerprint image correction model based on high dynamic range image fusion is obtained.

[0120] The core of the model lies in its fusion capability. It can intelligently analyze the input initial fingerprint image and the third fingerprint image, extract their respective edge intensity features and texture features, and adjust the contribution of these features in the fusion process through weight coefficients. Specifically, through a specific fusion formula, the feature sets of the two images are weighted and fused to obtain the fused edge and texture features. Subsequently, these fused features are used to reconstruct the image and generate the final result image. This process significantly improves the overall quality of the fingerprint image, increases the number of effective feature points, and retains the fingerprint's detail information and global illumination information, providing a more reliable basis for subsequent fingerprint reading.

[0121] In summary, when the third fingerprint image does not meet the feature point threshold, by introducing a fingerprint image correction model based on high dynamic range image fusion, the quality of the fingerprint image can be effectively improved, the number of effective feature points can be increased, and the accuracy and reliability of fingerprint recognition can be ensured. This process not only reflects the in-depth application of image processing technology in the field of fingerprint recognition, but also provides strong support for improving user experience and enhancing system security.

[0122] In one implementation, the training process of the fingerprint image correction model based on high dynamic range image fusion includes:

[0123] Building a generative adversarial network model based on the features of high dynamic range images and a multi-image fusion algorithm, and training the generative adversarial network model;

[0124] When the training times of the generative adversarial network model is greater than or equal to the preset maximum training times, or when the loss function value of the generative adversarial network model is less than the preset loss threshold, the training is determined to be completed, and a fingerprint image correction model based on high dynamic range image fusion is obtained.

[0125] It should be noted that this process involves the generative adversarial network (GANs) technology in deep learning, which is a powerful model for generating new data samples.

[0126] In the training process of the fingerprint image correction model based on high dynamic range (HDR) image fusion, first, a set of high dynamic range fingerprint images need to be collected. These images can capture a wider range of brightness and color details, thus providing richer information than standard images. Then, a multi-image fusion algorithm is used to enhance the image and extract features that are helpful for fingerprint correction by combining the features of HDR images (such as contrast, saturation, and exposure).

[0127] The architecture of the model, the generative adversarial network consists of two parts, namely the generator (G) and the discriminator (D). The goal of the generator is to generate fingerprint images that are close to the real ones, while the goal of the discriminator is to distinguish the generated images from the real ones. During the training process, the generator and the discriminator are trained in an adversarial manner. The generator continuously generates images, and the discriminator continuously tries to distinguish them. The abilities of the two promote each other in this process and make progress together.

[0128] The loss function is used during training to evaluate the performance of the model. The loss function is an indicator that evaluates the difference between the predicted output and the actual output. In generative adversarial networks, the discriminator is usually trained using the binary cross entropy loss function, while the generator is trained using a combination of the generator loss and the discriminator loss.

[0129] To avoid overfitting, the training process stops when the model's training times reach the preset maximum number of training times or the loss function value is less than the preset loss threshold. This early stopping mechanism helps the model maintain good generalization ability.

[0130] After training is complete, the model needs to be evaluated to ensure that it can accurately correct fingerprint images. Evaluation usually involves calculating the model's performance metrics such as precision, recall, and F1 score on an independent test set.

[0131] After getting the preliminary model, some optimization steps are needed to further improve the performance of the model. This includes hyperparameter tuning and model pruning.

[0132] Finally, the trained and optimized model can be deployed in the actual fingerprint recognition system.

[0133] In one embodiment, the method comprises:

[0134] Feature point detection is performed using the following formula:

[0135]

[0136] Where S(x, y) represents the intensity of the feature point at the image coordinate (x, y), I x and I y are the gradient values ​​of the image in the x and y directions respectively, and θ is the preset feature point threshold;

[0137] When S(x, y)>θ is satisfied, the point (x, y) is identified as a feature point;

[0138] When the counted number of feature points is less than a preset feature point number threshold, it is determined that the valid feature points included in the third fingerprint image do not meet the requirements.

[0139] It should be noted that the detection of feature points depends on the gradient information of the image and the preset threshold. The gradient value reflects the intensity and direction of the image brightness change, and is the basis for identifying key features in the image (such as edges, corners, and textures). In fingerprint images, these feature points usually correspond to the intersection of the ridges and valleys of the fingerprint, and are an important basis for fingerprint matching and recognition.

[0140] The gradient value can be calculated by Sobel, Prewitt or Canny operators, which can provide information about the edge direction and strength of the local area of ​​the image.

[0141] The formula for calculating the strength of feature points is a function based on the gradient value, which combines the size and direction information of the gradient to evaluate the possibility of each pixel as a feature point. The preset feature point threshold is a key parameter that determines which pixels have a high enough gradient strength to be considered as feature points. This threshold needs to be adjusted according to the specific application scenario and image characteristics to achieve the best detection effect.

[0142] When the gradient strength of a point meets the preset conditions, the point is identified as a feature point. If the number of feature points identified in the image does not reach the preset minimum number threshold, this may mean that the image quality is poor or the feature points are too sparse for effective recognition or matching. In this case, additional processing steps are required.

[0143] In one implementation, the step of inputting the initial fingerprint image and the third fingerprint image into a pre-trained fingerprint image correction model based on high dynamic range image fusion to obtain a final result image includes:

[0144] Extracting features from the initial fingerprint image and the third fingerprint image respectively to obtain a feature set including image edge intensity features and texture features;

[0145] Based on the first weight coefficient and the second weight coefficient, the feature set is fused to obtain a fused edge feature and a fused texture feature;

[0146] Based on the third weight coefficient and the fourth weight coefficient, image reconstruction is performed on the fused edge features and the fused texture features to obtain a final result image.

[0147] Exemplarily, the features of the initial fingerprint image are extracted to obtain a feature set of image edge intensity features and texture features of the initial fingerprint image; the features of the third fingerprint image are extracted to obtain a feature set of image edge intensity features and texture features of the third fingerprint image. Then, based on the first weight coefficient and the second weight coefficient, the feature set is fused. For example, the first weight coefficient is α, the second weight coefficient is 1-α, α is a weight coefficient between 0 and 1, the edge intensity features of the two images are fused based on the first weight coefficient and the second weight coefficient to obtain the fused edge features, and the texture features of the two images are fused based on the coefficient to obtain the fused texture features.

[0148] Exemplarily, the third weight coefficient is β, and the fourth weight coefficient is 1-β, where β is a weight coefficient used to adjust the contribution of edges and textures in the final image.

[0149] It should be noted that in this embodiment, features are first extracted from the two images, including edge intensity features and texture features. Edge features are usually composed of high-frequency information in the image, such as calculated by Sobel or Canny operators, while texture features describe the pattern and shape of the area in the image. These features are collected into a feature set, and then fused by a weighted average formula, where α is a weight coefficient for balancing the contribution of the two feature sets. This fusion strategy allows the model to select features that rely more on the initial image or the third image according to the specific situation, thereby generating a feature set that integrates the information of the two images.

[0150] In the image reconstruction stage, the fused edge and texture features are used to reconstruct the final image through another weighted formula combined with the weight coefficient β. This weight coefficient allows the model to adjust the relative importance of edges and textures in the reconstructed image, thereby generating a high-quality image that retains details and has clear textures.

[0151] This method is particularly useful when dealing with high dynamic range image fusion, because it can effectively balance the image features at different exposure levels to generate an image with a wider dynamic range. In addition, this technology can also reduce noise and artifacts in the image and improve the signal-to-noise ratio of the image, which is crucial for high-precision applications such as fingerprint recognition.

[0152] In order to facilitate the understanding of the present invention, some preferred embodiments of the present invention are further described below.

[0153] In one implementation, the present invention uses a fingerprint image correction model based on high dynamic range image fusion, which can handle the problem of inconsistent image quality caused by changes in ambient light or differences in imaging devices. Through this implementation, features can be extracted from images of multiple exposure levels and fused into a high-quality fingerprint image, thereby improving the accuracy and reliability of fingerprint recognition.

[0154] In another implementation, an adaptive threshold processing mechanism can also be introduced to dynamically adjust the parameters in the image processing process. This mechanism can automatically adjust the weight coefficients in the fusion algorithm and other related image processing parameters based on the real-time analysis results of the image. For example, if the image analysis results show that the contrast of the input image is low, the system can automatically increase the weight of edge enhancement to improve the contrast and recognizable features of the final image.

[0155] The following describes the working process of the present invention using a common scenario as an example. Figure 2 , which is Figure 1 Schematic diagram of the working scenario of the method.

[0156] When the user places his finger on the LCD fingerprint module, the sensor built into the module immediately captures the initial fingerprint image. Subsequently, the system performs preprocessing operations on the image, including noise reduction and contrast enhancement. The processed image is input into the LCD fingerprint reading processing model, which extracts the key feature points of the fingerprint through a complex network structure and training data. If the model determines that the extracted feature points do not meet the preset threshold requirements, the fingerprint image correction model based on HDR image fusion is further started to fuse the initial fingerprint image and the preliminarily processed image to generate a final result image containing more valid feature points. Finally, the system performs fingerprint reading and comparison operations based on the final result image to complete the entire recognition process.

[0157] In summary, the method of the present invention not only improves the accuracy of fingerprint recognition, but also enhances the robustness of the system, ensuring that fingerprint images can be accurately captured and processed even under undesirable imaging conditions.

[0158] Reference Figure 2 The second embodiment of the present invention provides a liquid crystal display fingerprint module reading system, comprising:

[0159] A data acquisition module is used to acquire an initial fingerprint image of a fingerprint module on a liquid crystal display;

[0160] A data processing module, used for performing a preprocessing operation on the initial fingerprint image to obtain a second fingerprint image;

[0161] A fingerprint reading processing module, used for inputting the second fingerprint image into a pre-trained liquid crystal display fingerprint reading processing model to obtain a third fingerprint image;

[0162] The fingerprint image optimization module is used to input the initial fingerprint image and the third fingerprint image into a pre-trained fingerprint image correction model based on high dynamic range image fusion to obtain a final result image when it is determined that the valid feature points contained in the third fingerprint image do not meet a preset feature point threshold, and perform fingerprint reading based on the final result image.

[0163] It should be noted that the LCD display fingerprint module reading system provided in an embodiment of the present invention is used to execute all the process steps of the LCD display fingerprint module reading method in the above embodiment. The working principles and beneficial effects of the two correspond one to one, so they will not be repeated here.

[0164] The embodiment of the present invention further provides an electronic device. The electronic device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a liquid crystal display fingerprint module reading program. When the processor executes the computer program, the steps in the above-mentioned liquid crystal display fingerprint module reading method embodiments are implemented, such as Figure 1 Alternatively, when the processor executes the computer program, the functions of each module / unit in the above-mentioned system embodiments are realized, such as the fingerprint reading processing module.

[0165] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, which are used to describe the execution process of the computer program in the electronic device.

[0166] The electronic device may be a computing device such as a desktop computer, a notebook, a PDA, and a smart tablet. The electronic device may include, but is not limited to, a processor and a memory. Those skilled in the art will appreciate that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. The electronic device may include more or fewer components than the above components, or a combination of certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0167] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the electronic device, and various interfaces and lines are used to connect various parts of the entire electronic device.

[0168] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the electronic device by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0169] Wherein, if the module / unit integrated in the electronic device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or system that can carry the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media does not include electrical carrier signals and telecommunication signals.

[0170] It should be noted that the system embodiment described above is merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the system embodiment provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0171] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for reading a fingerprint module on a liquid crystal display, characterized in that: include: Get the initial fingerprint image of the LCD fingerprint module; Performing a preprocessing operation on the initial fingerprint image to obtain a second fingerprint image; Inputting the second fingerprint image into a pre-trained LCD display fingerprint reading processing model to obtain a third fingerprint image; When it is determined that the valid feature points contained in the third fingerprint image do not meet the preset feature point threshold, the initial fingerprint image and the third fingerprint image are input into a pre-trained fingerprint image correction model based on high dynamic range image fusion to obtain a final result image, and fingerprint reading is performed based on the final result image.

2. The LCD fingerprint module reading method according to claim 1, characterized in that: The training process of the LCD display fingerprint reading processing model includes: Building a convolutional neural network model based on the grayscale, texture and edge features of the historical fingerprint image, and training the convolutional neural network model; When the training times of the convolutional neural network model is greater than or equal to the preset maximum training times, or when the loss function value of the convolutional neural network model is less than the preset loss threshold, the training is determined to be completed and a liquid crystal display fingerprint reading processing model is obtained.

3. The LCD fingerprint module reading method according to claim 1, characterized in that: The training process of the fingerprint image correction model based on high dynamic range image fusion includes: Building a generative adversarial network model based on the features of high dynamic range images and a multi-image fusion algorithm, and training the generative adversarial network model; When the training times of the generative adversarial network model is greater than or equal to the preset maximum training times, or when the loss function value of the generative adversarial network model is less than the preset loss threshold, the training is determined to be completed, and a fingerprint image correction model based on high dynamic range image fusion is obtained.

4. The LCD fingerprint module reading method according to claim 1, characterized in that: The preprocessing operation on the initial fingerprint image includes: Noise reduction is achieved through the following linear filtering formula: Where N is the total number of pixels in the filter window. For a (2a+1)×(2b+1) window, N=(2a+1)(2b+1), a and b are the radii of the filter window in the horizontal and vertical directions, respectively. m and n are the coordinate offsets relative to the center pixel (x, y). I′(x, y) is the pixel value at the coordinate (x, y) of the image after linear filtering and denoising. Perform Fourier transform on the denoised image: Among them, u and v are frequency variables, u represents the frequency of the image in the horizontal direction, v represents the frequency of the image in the vertical direction, j is the imaginary unit, It means Fourier transform of I′(x, y); Filter out high-frequency components of the Fourier transformed image: F′(u,v)=F(u,v)·L(u,v) Among them, L(u, v) is a low-pass filter, · represents the dot product operation, Where D is the cutoff frequency, Represents the distance from the origin of the frequency domain to the point (u, v); Perform inverse Fourier transform on the frequency domain image after filtering out the high-frequency components, and convert the image from the frequency domain back to the spatial domain: Where I″(x, y) is the pixel value of the denoised and smoothed image at coordinate (x, y), is the inverse Fourier transform.

5. The LCD fingerprint module reading method according to claim 1, characterized in that: The step of inputting the second fingerprint image into a pre-trained liquid crystal display fingerprint reading processing model to obtain a third fingerprint image includes: Performing image segmentation and threshold processing on the second fingerprint image to obtain an enhanced fingerprint image; The enhanced fingerprint image is thinned and integrated to obtain the third fingerprint image.

6. The LCD fingerprint module reading method according to claim 1, characterized in that: The method comprises: Feature point detection is performed using the following formula: Where S(x, y) represents the intensity of the feature point at the image coordinate (x, y), I x and I y are the gradient values ​​of the image in the x and y directions respectively, and θ is the preset feature point threshold; When S(x, y)>θ is satisfied, the point (x, y) is identified as a feature point; When the counted number of feature points is less than a preset feature point number threshold, it is determined that the valid feature points included in the third fingerprint image do not meet the requirements.

7. The method for reading fingerprint module of liquid crystal display according to claim 2, characterized in that: The step of inputting the initial fingerprint image and the third fingerprint image into a pre-trained fingerprint image correction model based on high dynamic range image fusion to obtain a final result image includes: Extracting features from the initial fingerprint image and the third fingerprint image respectively to obtain a feature set including image edge intensity features and texture features; Based on the first weight coefficient and the second weight coefficient, the feature set is fused to obtain a fused edge feature and a fused texture feature; Based on the third weight coefficient and the fourth weight coefficient, image reconstruction is performed on the fused edge features and the fused texture features to obtain a final result image.

8. A liquid crystal display fingerprint module reading system, characterized in that: include: A data acquisition module is used to acquire an initial fingerprint image of a fingerprint module on a liquid crystal display; A data processing module, used for performing a preprocessing operation on the initial fingerprint image to obtain a second fingerprint image; A fingerprint reading processing module, used for inputting the second fingerprint image into a pre-trained liquid crystal display fingerprint reading processing model to obtain a third fingerprint image; The fingerprint image optimization module is used to input the initial fingerprint image and the third fingerprint image into a pre-trained fingerprint image correction model based on high dynamic range image fusion to obtain a final result image when it is determined that the valid feature points contained in the third fingerprint image do not meet a preset feature point threshold, and perform fingerprint reading based on the final result image.

9. An electronic device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the liquid crystal display fingerprint module reading method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the liquid crystal display fingerprint module reading method according to any one of claims 1 to 7.

Images