Intelligent door lock control method and device, intelligent door lock and computer readable medium

Abstract

The embodiment of the invention discloses an intelligent door lock control method and device, an intelligent door lock and a computer readable medium. A specific implementation mode of the method comprises the following steps: acquiring a palm vein image; inputting the palm vein image into a compressed sampling layer to obtain a compressed sensing measurement vector; inputting the compressed sensing measurement vector into an initial reconstruction image layer to obtain an initial reconstruction image; inputting the initial reconstruction image into a depth reconstruction image layer to obtain a depth reconstruction image; inputting the depth reconstruction image into a global local feature reconstruction image layer to obtain a palm vein reconstruction image; performing region-of-interest extraction on the palm vein reconstruction image to obtain a region-of-interest image; performing feature extraction on the region-of-interest image to obtain a region-of-interest feature vector; generating a palm vein matching result; and in response to determining that the palm vein matching result represents successful matching, controlling the door lock assembly to execute unlocking operation. According to the embodiment, waste of collected image resources can be reduced.

Description

Technical Field

[0001] Embodiments of the present disclosure relate to the field of computer technologies, and particularly to an intelligent door lock control method, apparatus, intelligent door lock, and computer-readable medium. Background Art

[0002] With the progress of technology, security has entered the intelligent era. As one of the important branches of security, intelligent door locks have brought more convenience to users' lives. Currently, when performing control operations on an intelligent door lock, the commonly used method is to capture an image of a target object (such as a palm or a face), perform related processing (such as binarization processing and denoising processing) on the image, extract features, and determine whether it is the target user through feature matching.

[0003] However, the inventors have found that when performing control operations on an intelligent door lock in the above manner, the following technical problems often exist:

[0004] First, after capturing the image, no consideration is given to performing image reconstruction operations on the captured image to improve the clarity of the image. As a result, the clarity of the image is low, which leads to low accuracy in image feature matching, requires multiple image captures, causing waste of image capture resources; at the same time, it causes inconvenience to users and poor user experience.

[0005] Second, during the process of image processing, no consideration is given to the uneven illumination of different parts of the captured image, and a preset constant threshold is used for binarization processing, resulting in poor clarity of the image after binarization processing, which leads to low accuracy in image feature matching, requires multiple image captures, causing waste of image capture resources; at the same time, it causes inconvenience to users and poor user experience.

[0006] Third, during the process of image denoising, the same method is used to perform denoising processing on all regions of the image, resulting in more loss of edge detail features in the region of interest of the image, low clarity of the image, which leads to low accuracy in image feature matching, requires multiple image captures, causing waste of image capture resources; at the same time, it causes inconvenience to users and poor user experience.

[0007] The above information disclosed in this background art section is only used to enhance the understanding of the background of the inventive concept, and thus, it may include information that does not form the prior art known to ordinary technicians in the art of this country. Summary of the Invention

[0008] The content of the present disclosure is partially used to introduce concepts in a brief form, which will be described in detail in the following detailed implementation section. The content of the present disclosure is not intended to identify the key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0009] Some embodiments of the present disclosure propose an intelligent door lock control method, a computer-readable medium, and a computer program product to solve one or more of the technical problems mentioned in the above background art section.

[0010] In a first aspect, some embodiments of the present disclosure provide an intelligent door lock control method, the method comprising: collecting a palm vein image of a user through a palm vein image acquisition device; inputting the palm vein image into a compressive sampling layer of a pre-trained image reconstruction model to obtain a compressive sensing measurement vector, wherein the image reconstruction model further comprises: an initial reconstructed image layer, a deep reconstructed image layer, and a global-local feature reconstructed image layer; inputting the compressive sensing measurement vector into the initial reconstructed image layer to obtain an initial reconstructed image; inputting the initial reconstructed image into the deep reconstructed image layer to obtain a deep reconstructed image; inputting the deep reconstructed image into the global-local feature reconstructed image layer to obtain a global-local feature reconstructed image as a palm vein reconstructed image; extracting a region of interest from the palm vein reconstructed image to obtain a region of interest image; extracting features from the region of interest image to obtain a region of interest feature vector; generating a palm vein matching result according to the region of interest feature vector and a preset feature vector set; and in response to determining that the palm vein matching result indicates a successful match, controlling a door lock component of the intelligent door lock to perform an unlocking operation.

[0011] Second aspect, some embodiments of the present disclosure provide an intelligent door lock control device, the device comprising: an acquisition unit configured to acquire a palm vein image of a user through a palm vein image acquisition device; a first input unit configured to input the palm vein image into a compressive sampling layer of a pre-trained image reconstruction model to obtain a compressive sensing measurement vector, wherein the image reconstruction model further comprises: an initial reconstructed image layer, a deep reconstructed image layer, and a global-local feature reconstructed image layer; a second input unit configured to input the compressive sensing measurement vector into the initial reconstructed image layer to obtain an initial reconstructed image; a third input unit configured to input the initial reconstructed image into the deep reconstructed image layer to obtain a deep reconstructed image; a fourth input unit configured to input the deep reconstructed image into the global-local feature reconstructed image layer to obtain a global-local feature reconstructed image as a palm vein reconstructed image; a region of interest extraction unit configured to perform region of interest extraction on the palm vein reconstructed image to obtain a region of interest image; a feature extraction unit configured to perform feature extraction on the region of interest image to obtain a region of interest feature vector; a generation unit configured to generate a palm vein matching result according to the region of interest feature vector and a preset feature vector set; and a control unit configured to control a door lock assembly of the intelligent door lock to perform an unlocking operation in response to determining that the palm vein matching result indicates a successful match.

[0012] Third aspect, some embodiments of the present disclosure provide an intelligent door lock, comprising: one or more processors; a door lock assembly; a storage device having stored thereon one or more programs, which when executed by the one or more processors cause the one or more processors to implement the method described in any implementation manner of the first aspect above.

[0013] Fourth aspect, some embodiments of the present disclosure provide a computer-readable medium having stored thereon a computer program, wherein when the program is executed by a processor, the method described in any implementation manner of the first aspect above is implemented.

[0014] The above-mentioned various embodiments of the present disclosure have the following beneficial effects: Through the intelligent door lock control method of some embodiments of the present disclosure, the clarity of the image to be matched can be improved, the accuracy of image feature matching can be increased, the waste of collected image resources can be reduced, and the user experience can be enhanced. Specifically, the reasons for the low clarity of the image, resulting in low accuracy of image feature matching, the need to collect images multiple times, and the waste of collected image resources; at the same time, it causes inconvenience to the user and a poor user experience are as follows: After the image is collected, no image reconstruction operation is considered for the collected image to improve the clarity of the image. The low clarity of the image leads to low accuracy of image feature matching, the need to collect images multiple times, and the waste of collected image resources; at the same time, it causes inconvenience to the user and a poor user experience. Based on this, in the intelligent door lock control method of some embodiments of the present disclosure, first, a palmar vein image of the user is collected by a palmar vein image acquisition device. Thus, a palmar vein image can be obtained, which can be used for matching and recognition. Then, the above palmar vein image is input into the compressive sampling layer of a pre-trained image reconstruction model to obtain a compressive sensing measurement vector. Among them, the above image reconstruction model further includes: an initial reconstructed image layer, a deep reconstructed image layer, and a global-local feature reconstructed image layer. Thus, a compressive sensing measurement vector after compression of the palmar vein image can be obtained. After that, the above compressive sensing measurement vector is input into the above initial reconstructed image layer to obtain an initial reconstructed image. Thus, through pixel shuffling operation, an initial reconstructed image can be obtained. Subsequently, the above initial reconstructed image is input into the above deep reconstructed image layer to obtain a deep reconstructed image. Thus, a deep reconstructed image can be obtained. Thus, through the image reconstruction operation, the clarity of the palmar vein image can be further improved. Second, the above deep reconstructed image is input into the above global-local feature reconstructed image layer to obtain a global-local feature reconstructed image as the palmar vein reconstructed image. Thus, a palmar vein reconstructed image can be obtained. Thus, the palmar vein image can be reconstructed from two branches of global features and local features and fused, further improving the clarity of the constructed image. Then, region of interest extraction is performed on the above palmar vein reconstructed image to obtain a region of interest image. Thus, a region of interest image representing the target region can be obtained. Thus, the region of the palmar vein image to be processed can be reduced, and the efficiency of processing the image can be improved. After that, feature extraction is performed on the above region of interest image to obtain a region of interest feature vector. Thus, a region of interest feature vector representing the palmar vein image can be obtained. Subsequently, a palmar vein matching result is generated according to the region of interest feature vector and a preset feature vector set. Thus, a palmar vein matching result can be obtained, which can be used to determine whether the user is an authorized user. Finally, in response to determining that the above palmar vein matching result indicates a successful match, the door lock component of the above intelligent door lock is controlled to perform an unlocking operation. Thus, according to the matching result, the door lock can be controlled to perform an unlocking operation. Thus, the user experience is improved.Also, after collecting the user's palm vein image, depth reconstruction processing is performed on the palm vein image through Fourier convolution operation, which improves the clarity of the reconstructed image. It is also because the palm vein image is reconstructed from two branches of global features and local features and fused, which further improves the clarity of the constructed image. Thereby, the accuracy of image feature matching is improved, the number of image acquisitions is reduced, and further, the waste of acquired image resources is reduced, and the user experience is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In combination with the accompanying drawings and with reference to the following specific embodiments, the above and other features, advantages, and aspects of the various embodiments of the present disclosure will become more apparent. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the elements and elements are not necessarily drawn to scale.

[0016] Figure 1 is a flowchart of some embodiments of the intelligent door lock control method according to the present disclosure;

[0017] Figure 2 is a schematic structural diagram of some embodiments of the intelligent door lock control device according to the present disclosure;

[0018] Figure 3 is a schematic structural diagram of an intelligent door lock suitable for implementing some embodiments of the present disclosure. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] The embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although some embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. On the contrary, these embodiments are provided to more thoroughly and completely understand the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not used to limit the protection scope of the present disclosure.

[0020] In addition, it should be noted that, for the sake of convenience of description, only the parts related to the relevant invention are shown in the drawings. Without conflict, the embodiments in the present disclosure and the features in the embodiments can be combined with each other.

[0021] It should be noted that the concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules, or units, and are not used to limit the order of functions performed by these devices, modules, or units or their interdependent relationships.

[0022] It should be noted that the modifications of "one" and "plural" mentioned in the present disclosure are illustrative rather than restrictive. Those skilled in the art should understand that unless otherwise clearly specified in the context, it should be understood as "one or more".

[0023] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are for illustrative purposes only and are not intended to limit the scope of these messages or information.

[0024] The present disclosure will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0025] Figure 1 Flow 100 of some embodiments of an intelligent door lock control method according to the present disclosure is shown. The intelligent door lock control method includes the following steps:

[0026] Step 101, collecting a palm vein image of a user through a palm vein image acquisition device.

[0027] In some embodiments, the execution subject (such as a computing device) of the intelligent door lock control method can collect a palm vein image of a user from the palm vein image acquisition device through a wired connection method or a wireless connection method. Among them, the above palm vein image acquisition device can be a device capable of collecting a palm vein image of a user. For example, the above palm vein image acquisition device can be a near-infrared camera. The acquisition parameters of the above palm vein image acquisition device can be: wavelength of 850 nm, resolution of 1920*1080, and lens angle of 60°. The above user can be any user. The above palm vein image is the palm vein image of the above user. It should be noted that the above wireless connection method can include but is not limited to 3G / 4G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other currently known or future-developed wireless connection methods.

[0028] Step 102, inputting the palm vein image into the compressive sampling layer of a pre-trained image reconstruction model to obtain a compressive sensing measurement vector.

[0029] In some embodiments, the above-mentioned execution entity may input the palm vein image into the compressive sampling layer of a pre-trained image reconstruction model to obtain a compressive sensing measurement vector. Among them, the above-mentioned image reconstruction model may further include: an initial reconstructed image layer, a deep reconstructed image layer, and a global-local feature reconstructed image layer. The above-mentioned image reconstruction model may be a neural network model that takes the palm vein image as input and the palm vein reconstructed image as output. The above-mentioned compressive sampling layer may be a network layer that performs convolutional compressive sampling on the above-mentioned palm vein image with each row element of a preset measurement matrix as a convolution kernel to obtain a compressive sensing measurement vector. Among them, the above-mentioned preset measurement matrix may be a matrix preset for subsampling the original signal in the image to obtain a compressive sensing measurement vector of the observable image. The above-mentioned preset measurement matrix may include, but is not limited to, any one of the following: a local Hadamard measurement matrix, a local Fourier measurement matrix, a sparse random measurement matrix, a Gaussian measurement matrix, and a Bernoulli matrix. The above-mentioned compressive sensing measurement vector may be a vector of a low-dimensional space vector after compressive sampling of the palm vein image. The above-mentioned compressive sampling layer may be a network layer that performs convolutional compressive sampling on the above-mentioned palm vein image with each row element of a preset measurement matrix as a convolution kernel to obtain a compressive sensing measurement vector. Among them, the above-mentioned predicted measurement vector may include, but is not limited to: a Gaussian measurement matrix and a Bernoulli matrix. The above-mentioned initial reconstructed image layer may be a network layer that uses a preset measurement matrix to perform transformation processing on the compressive sensing measurement vector to obtain a reconstructed image corresponding to the palm vein image. The above-mentioned deep reconstructed image layer may be a network layer with a global receptive field and cross-scale feature fusion within the convolutional unit. Specifically, it is a network layer that linearly combines the pixels in the local neighborhood of the input image and then reconstructs the pixels at the corresponding positions in the output image. The above-mentioned global-local feature reconstructed image layer may be a network layer that uses a dual-branch network to separately extract global feature information and local feature information and perform feature fusion to obtain a reconstructed image.

[0030] Optionally, before the above-mentioned step of inputting the above-mentioned palm vein image into the compressive sampling layer of the pre-trained image reconstruction model to obtain a compressive sensing measurement vector, the above-mentioned execution entity may further perform the following steps:

[0031] First step, perform grayscale transformation processing on the above-mentioned palm vein image to obtain the palm vein image after grayscale transformation processing as the grayscale-transformed palm vein image. In practice, the above-mentioned execution entity may perform grayscale transformation processing on the above-mentioned palm vein image through the following steps to obtain the palm vein image after grayscale transformation processing as the grayscale-transformed palm vein image:

[0032] First sub-step, determine the maximum grayscale value among the respective grayscale values corresponding to each pixel point included in the above-mentioned palm vein image as the maximum grayscale value.

[0033] The second sub-step is to, for each gray value corresponding to the above-mentioned palm vein image, determine the absolute value of the difference between the above-mentioned maximum gray value and the above-mentioned gray value as the inverted gray value.

[0034] The third sub-step is to determine the determined inverted gray values as an inverted gray value set.

[0035] The fourth sub-step is to determine the image composed of the above-mentioned inverted gray value set as the gray-scale transformed palm vein image.

[0036] The second step is to perform image dilation processing on the above-mentioned gray-scale transformed palm vein image according to a preset structural element set to generate a set of dilated palm vein images. Among them, the sizes of the various preset structural elements in the above-mentioned preset structural element set are different. Here, the preset structural elements in the above-mentioned preset structural element set can be preset circular structural elements. The radii corresponding to the various preset structural elements included in the above-mentioned preset structural element set can be 1, 2, 3, 4, 5, and 6. The unit of the above-mentioned radius is pixels. As an example, the dilated palm vein image in the set of dilated palm vein images can be expressed by the following formula:

[0037]

[0038] Among them, I2(i, j) can represent the output image, that is, the dilated palm vein image. The above-mentioned i can represent the abscissa of the output / input image. The above-mentioned j can represent the ordinate of the output / input image. I1 can represent the input image, that is, the gray-scale transformed palm vein image. can represent dilation processing. n can represent the scale (the radius of the corresponding structural element is 1, 2, 3, 4, 5, 6). K can represent the structural element. The above-mentioned x can represent the abscissa of the structural element. The above-mentioned y can represent the ordinate of the structural element. D b can represent the domain of definition of the input image. i - x can represent the abscissa value in the input image corresponding to the abscissa x of the structural element. j - y can represent the ordinate value in the input image corresponding to the ordinate y of the structural element. I1(i - x, j - y) can represent the image gray value at the position corresponding to the structural element in the input image. max{I1(i - x, j - y) + nK(x, y)|(x, y) ∈ D b} can represent the maximum gray value in the area corresponding to the structural element after dilation processing.

[0039] The third step is to generate a set of eroded palm vein images according to the above-mentioned preset structural element set and the above-mentioned set of dilated palm vein images. Among them, the eroded palm vein images in the above-mentioned set of eroded palm vein images correspond one by one to the dilated palm vein images in the above-mentioned set of dilated palm vein images. As an example, the eroded palm vein images in the above-mentioned set of eroded palm vein images can be expressed by the following formula:

[0040]

[0041] Among them, I3(i, j) can represent the eroded palm vein image. I2 can represent the dilated palm vein image. can represent the erosion operation. I2(i - x, j - y) can represent the image gray value at the corresponding position of the structural element in the dilated palm vein image. D c can represent the domain of the dilated palm vein image. min{I2(i - x, j - y) - nK(x, j)|(x, y) ∈ D c} can represent the minimum gray value in the area corresponding to the structural element after erosion processing.

[0042] Step 4: Generate an image standard deviation information set corresponding to the above eroded palm vein image set according to the above eroded palm vein image set and the above gray - transformed palm vein image. As an example, the image standard deviation information in the above image standard deviation information set can be expressed by the following formula:

[0043]

[0044] Among them, the above σ i can represent the image standard deviation information in the image standard deviation information set. m can represent the length of the image. n can represent the width of the image. i can represent the serial number of each eroded palm vein image in the eroded palm vein image set. I3(x, y) can represent the eroded palm vein image in the eroded palm vein image set. I1(x, y) can represent the gray - transformed palm vein image.

[0045] Step 5: Determine a weight information set corresponding to the above eroded palm vein image set according to the above image standard deviation information set. As an example, the weight information in the above weight information set can be expressed by the following formula:

[0046]

[0047] Among them, the above λ i can represent the weight information in the weight information set. σ j can represent the image standard deviation information in the image standard deviation information set. σ i can represent the image standard deviation information corresponding to the above weight information. n can represent the number of each image standard deviation information included in the image standard deviation information set. j can represent the serial number of the image standard deviation information in the image standard deviation information set. The above can represent the sum of each image standard deviation information included in the image standard deviation information set.

[0048] Step 6: Generate a fused palm vein image according to the above weight information set and the above eroded palm vein image set. In practice, first, for each eroded palm vein image in the above eroded palm vein image set, the execution entity can determine the product of the eroded palm vein image and the weight information corresponding to the eroded palm vein image in the above weight information set as a weighted palm vein image. Then, the determined weighted palm vein images can be superimposed to obtain the superimposed weighted palm vein images as the fused palm vein image. As an example, the above fused palm vein image can be expressed by the following formula:

[0049]

[0050] wherein, the above I4(x, y) can represent the fused palm vein image. λ i can represent the weight information in the weight information set. I3(i, j) can represent the eroded palm vein image in the eroded palm vein image set. n can represent the number of eroded palm vein images included in the eroded palm vein image set. i can represent the serial number of the eroded palm vein image in the eroded palm vein image set.

[0051] Step 7: Determine the above fused palm vein image as the palm vein image to update the above palm vein image.

[0052] The above first step to the seventh step and their related content are an inventive point of the embodiments of the present disclosure, which solves the third technical problem mentioned in the background art: "During the process of denoising an image, the entire image area is uniformly denoised (using the same method), resulting in a large loss of edge detail features in the region of interest of the image, low clarity of the image, thus leading to low accuracy in image feature matching, requiring multiple acquisitions of images, causing waste of image acquisition resources; at the same time, it causes inconvenience to users and poor user experience." The factors that result in a large loss of edge detail features in the region of interest of the image, low clarity of the image, thus leading to low accuracy in image feature matching, requiring multiple acquisitions of images, causing waste of image acquisition resources; at the same time, it causes inconvenience to users and poor user experience are often as follows: During the process of denoising an image, the entire image area is uniformly denoised (using the same method), resulting in a large loss of edge detail features in the region of interest of the image, low clarity of the image, thus leading to low accuracy in image feature matching, requiring multiple acquisitions of images, causing waste of image acquisition resources; at the same time, it causes inconvenience to users and poor user experience. If the above factors are solved, the clarity of the image to be matched can be improved, the accuracy of image feature matching can be increased, the waste of image acquisition resources can be reduced, and the user experience can be enhanced. To achieve this effect, first, perform gray-scale transformation processing on the above palm vein image to obtain the palm vein image after gray-scale transformation processing as the gray-scale transformed palm vein image. Thus, the gray-scale transformed palm vein image can be obtained, which can be used to enhance the contrast between the target features in the region of interest and the background. Then, according to the preset structural element set, perform image dilation processing on the above gray-scale transformed palm vein image to generate a set of dilated palm vein images. Among them, the sizes of the various preset structural elements in the above preset structural element set are different. Thus, a set of dilated palm vein images can be obtained. According to the above preset structural element set and the above set of dilated palm vein images, generate a set of eroded palm vein images. Among them, the eroded palm vein images in the above set of eroded palm vein images correspond to the dilated palm vein images in the above set of dilated palm vein images. Thus, a set of eroded palm vein images can be obtained. Thereby, noise can be suppressed by large-scale preset structural elements, and detail features can be retained by small-scale preset structural elements. Secondly, according to the above set of eroded palm vein images and the above gray-scale transformed palm vein image, generate an image standard deviation information set corresponding to the above set of eroded palm vein images. Thus, an image standard deviation information set can be obtained, which can be used to measure the denoising effect of the image. Then, according to the above image standard deviation information set, determine a weight information set corresponding to the above set of eroded palm vein images. Thus, a weight information set representing the contribution degree of the denoising effect can be obtained. Finally, according to the above weight information set and the above set of eroded palm vein images, generate a fused palm vein image.Determine the above-mentioned fused palm vein image as the palm vein image to update the above-mentioned palm vein image. Thus, a fused palm vein image representing the overall denoising effect can be obtained, that is, the palm vein image. Also, because the palm vein image is denoised by using a preset set of structural elements representing different scales, noise can be suppressed and more edge detail features of the regions of interest can be retained. Additionally, because the denoising contribution degrees of the preset structural elements of different scales to the palm vein image can be used to set denoising weights, the denoising effect is further enhanced. Thereby, the clarity of the image is improved, the accuracy of image feature matching is increased, the waste of acquired image resources is reduced, and the user experience is improved.

[0053] Step 103: Input the compressed sensing measurement vector into the initial reconstruction image layer to obtain an initial reconstruction image.

[0054] In some embodiments, the above-mentioned execution entity may input the above-mentioned compressed sensing measurement vector into the above-mentioned initial reconstruction image layer to obtain an initial reconstruction image. In practice, the above-mentioned execution entity may input the above-mentioned compressed sensing measurement vector into the above-mentioned initial reconstruction image layer to obtain an initial reconstruction image through the following steps:

[0055] The first step: Perform convolution processing on the above-mentioned compressed sensing measurement vector to obtain an initial reconstruction convolution vector. Here, convolution processing may be performed with the transpose matrix of a preset measurement matrix as the convolution kernel. The size of the transpose matrix of the above-mentioned preset measurement matrix may be the same as the data dimension size of the above-mentioned compressed sensing measurement vector.

[0056] The second step: Perform transformation processing on the above-mentioned initial reconstruction convolution vector to generate an initial reconstruction image. In practice, the above-mentioned execution entity may perform transformation processing on the above-mentioned initial reconstruction convolution vector through a pixel shuffling operation to generate an initial reconstruction image.

[0057] Step 104: Input the initial reconstruction image into the depth reconstruction image layer to obtain a depth reconstruction image.

[0058] In some embodiments, the above-mentioned execution entity may input the above-mentioned initial reconstruction image into the above-mentioned depth reconstruction image layer to obtain a depth reconstruction image. Among them, the above-mentioned depth reconstruction image layer may include: 1 layer of fast Fourier convolution, 5 fast Fourier convolution residual modules, and a 1*1 convolution layer. The above-mentioned fast Fourier convolution may be used to extract global context information. The size of the convolution kernel corresponding to the above-mentioned fast Fourier convolution may be 3*3*64.

[0059] Optionally, after the above-mentioned input of the above-mentioned initial reconstruction image into the above-mentioned depth reconstruction image layer to obtain a depth reconstruction image, the above-mentioned execution entity may further perform the following steps:

[0060] In the first step, the above-mentioned initial reconstructed image and the depth reconstructed image are combined to generate a combined depth reconstructed image. In practice, the above-mentioned execution entity can superimpose the above-mentioned initial reconstructed image and the depth reconstructed image to generate a combined depth reconstructed image. Here, the superimposing method can be to add the corresponding pixels of the above-mentioned initial reconstructed image and the depth reconstructed image.

[0061] In the second step, the combined depth reconstructed image is determined as the depth reconstructed image to update the depth reconstructed image.

[0062] Step 105: Input the depth reconstructed image into the global-local feature reconstructed image layer to obtain the global-local feature reconstructed image as the palm vein reconstruction image.

[0063] In some embodiments, the above-mentioned execution entity can input the above-mentioned depth reconstructed image into the above-mentioned global-local feature reconstructed image layer to obtain the global-local feature reconstructed image as the palm vein reconstruction image. Among them, the above-mentioned global-local feature reconstructed image layer can be a two-branch image layer including a global feature reconstructed image network and a local feature reconstructed image network. In practice, the above-mentioned execution entity can input the above-mentioned depth reconstructed image into the above-mentioned global-local feature reconstructed image layer through the following steps to obtain the global-local feature reconstructed image as the palm vein reconstruction image:

[0064] In the first step, input the above-mentioned depth reconstructed image into the above-mentioned global feature reconstructed image network to obtain the global feature reconstructed image. Among them, the above-mentioned global feature reconstructed image network is a convolutional network with a large receptive field and a relatively deep network depth.

[0065] In the second step, input the above-mentioned depth reconstructed image into the above-mentioned local feature reconstructed image network to obtain the local feature reconstructed image. The above-mentioned local feature reconstructed image network can be a convolutional network with full resolution but a relatively shallow network depth.

[0066] In the third step, superimpose the above-mentioned global feature reconstructed image and the above-mentioned local feature reconstructed image to obtain a combined depth reconstructed image. Here, the superimposing method can be to add the corresponding elements of the above-mentioned global feature reconstructed image and the above-mentioned local feature reconstructed image.

[0067] In the fourth step, fuse the above-mentioned combined depth reconstructed image and the above-mentioned depth reconstructed image to generate the palm vein reconstruction image. Here, the fusing method can be to multiply the corresponding pixels of the above-mentioned combined depth reconstructed image and the above-mentioned depth reconstructed image.

[0068] Step 106: Extract the region of interest from the palm vein reconstruction image to obtain the region of interest image.

[0069] In some embodiments, the above-mentioned execution entity may extract a region of interest from the above-mentioned palm vein reconstruction image to obtain a region of interest image.

[0070] In some alternative implementation manners of some embodiments, the above-mentioned execution entity may extract a region of interest from the above-mentioned palm vein reconstruction image through the following steps to obtain a region of interest image:

[0071] First step, according to the above-mentioned palm vein reconstruction image and a pre-trained palm key point detection model, generate palm key point information. Among them, the above-mentioned palm key point information may include: first valley bottom key point information, second valley bottom key point information, and wrist center key point information. The above-mentioned first valley bottom key point information is the coordinate information of the valley bottom between the index finger and the middle finger. The above-mentioned second valley bottom key point information is the coordinate information of the valley bottom between the middle finger and the ring finger. The above-mentioned palm key point detection model may be a neural network model that takes the palm vein reconstruction image as the input and the palm key points as the output. For example, the above-mentioned palm key point detection model may be a ShuffleNet network model.

[0072] Second step, determine the image rotation angle according to the above-mentioned first valley bottom key point information and the above-mentioned second valley bottom key point information.

[0073] Third step, perform a rotation process on the above-mentioned preprocessed palm vein image according to the above-mentioned rotation angle to generate a rotation-processed image. In practice, the above-mentioned execution entity may rotate the above-mentioned preprocessed palm vein image in the clockwise direction by the corresponding rotation angle to generate a rotation-processed image.

[0074] Fourth step, determine the first and second valley bottom midpoint information according to the above-mentioned first valley bottom key point information and the above-mentioned second valley bottom key point information. In practice, first, the above-mentioned execution entity may determine the abscissa of the first and second valley bottom midpoint information as the average value of the abscissa of the above-mentioned first valley bottom key point information and the abscissa of the above-mentioned second valley bottom key point information. Then, the ordinate of the first and second valley bottom midpoint information may be determined as the average value of the ordinate of the above-mentioned first valley bottom key point information and the ordinate of the above-mentioned second valley bottom key point information. Finally, the coordinate information composed of the abscissa and ordinate of the above-mentioned first and second valley bottom midpoint information is determined as the first and second valley bottom midpoint information.

[0075] Step 5: Determine the palm center coordinate information based on the above information of the midpoints of the first and second valley bottoms and the above information of the key point of the wrist center. In practice, first, the above-mentioned execution entity can determine the abscissa of the palm center coordinate information as the average value of the abscissa of the above information of the first and second valley bottom midpoints and the abscissa of the above information of the key point of the wrist center. Then, the ordinate of the palm center coordinate information can be determined as the average value of the ordinate of the above information of the first and second valley bottom midpoints and the ordinate of the above information of the key point of the wrist center. Finally, the information composed of the abscissa and ordinate of the palm center coordinate information can be determined as the palm center coordinate information.

[0076] Step 6: Determine the image of the region of interest of the above user based on the above palm center coordinate information and the above rotation-processed image. In practice, the above-mentioned execution entity can use the palm center coordinate information as the center point of the rotation-processed image, use the preset image width as the image width of the region of interest image, and use the preset image height as the image height of the region of interest image for image cropping processing to obtain the rotation-processed image after image cropping processing as the region of interest image. Among them, the above preset image width can be the pre-set image width. The above preset image height can be the pre-set image height. For example, the above preset image width can be 3 cm. The above preset image height can be 3 cm.

[0077] In some optional implementation manners of some embodiments, based on the above information of the first valley bottom key point and the second valley bottom key point, the above-mentioned execution entity can determine the image rotation angle through the following steps:

[0078] Step 1: Determine the abscissa difference information as the difference between the abscissa included in the above information of the first valley bottom key point and the abscissa included in the above information of the second valley bottom key point.

[0079] Step 2: Determine the ordinate difference information as the difference between the ordinate included in the above information of the first valley bottom key point and the ordinate included in the above information of the second valley bottom key point.

[0080] Step 3: Determine the coordinate difference information as the ratio of the above abscissa difference information to the above ordinate difference information.

[0081] Step 4: Determine the angle corresponding to the arctangent of the above coordinate difference information as the image rotation angle.

[0082] Step 107: Extract features from the image of the region of interest to obtain the feature vector of the region of interest.

[0083] In some embodiments, the above-mentioned execution entity can extract features from the above image of the region of interest to obtain the feature vector of the region of interest.

[0084] In practice, the above-mentioned execution entity can use a feature extraction algorithm to extract features from the above-mentioned region of interest image, obtaining a feature vector of the region of interest. Among them, the above-mentioned feature extraction algorithm includes but is not limited to: SIFT (Scale-invariant feature transform) algorithm, SURF (Speeded Up Robust Features) algorithm, and Affine-SIFT (Affine-Scale-invariant feature transform, ASIFT).

[0085] Optionally, before the above-mentioned feature extraction from the above-mentioned region of interest image to obtain a feature vector of the region of interest, the above-mentioned execution entity can also perform the following steps:

[0086] First step, perform binarization processing on the above-mentioned region of interest image to obtain the binarized region of interest image as the binarized region of interest image.

[0087] Second step, perform contrast stretching processing on the above-mentioned binarized region of interest image to obtain the contrast-stretched binarized region of interest image as the contrast-stretched processed image. In practice, the above-mentioned execution entity can perform contrast stretching processing on the binarized region of interest image through CLAHE (Contrast Limited Adaptive Histogram Equalization) to obtain the contrast-stretched binarized region of interest image as the contrast-stretched processed image.

[0088] Third step, perform image enhancement processing on the above-mentioned contrast-stretched processed image to generate an enhanced region of interest image. Among them, the above-mentioned image enhancement processing can include but is not limited to: gray normalization processing and histogram equalization processing. In practice, the above-mentioned execution entity can perform histogram equalization processing on the above-mentioned contrast-stretched processed image to generate the contrast-stretched processed image after histogram equalization processing as the enhanced region of interest image.

[0089] Fourth step, determine the above-mentioned enhanced region of interest image as the region of interest image to update the above-mentioned region of interest image.

[0090] In some optional implementation manners of some embodiments, the above-mentioned execution entity can perform binarization processing on the above-mentioned region of interest image through the following steps to obtain the binarized region of interest image as the binarized region of interest image:

[0091] Step 1: Determine the center point coordinates of the above-mentioned region of interest image. In practice, the above-mentioned execution entity may determine the center point coordinates of the horizontal and vertical coordinates of the above-mentioned region of interest image as the center point coordinates of the above-mentioned region of interest image.

[0092] Step 2: Determine the set of sub-region of interest images according to the above-mentioned center point coordinates and the above-mentioned region of interest image. In practice, the above-mentioned execution entity may use the above-mentioned center point coordinates as the segmentation origin, use the horizontal axis corresponding to the segmentation origin as the horizontal segmentation line, and use the vertical axis corresponding to the segmentation origin as the vertical segmentation line to evenly divide the above-mentioned region of interest image into four sub-region of interest images as the set of sub-region of interest images. The above-mentioned set of sub-region of interest images may include: the first sub-region of interest image, the second sub-region of interest image, the third sub-region of interest image, and the fourth sub-region of interest image. The above-mentioned first sub-region of interest image and the above-mentioned third sub-region of interest image are diagonal images. The above-mentioned second sub-region of interest image and the above-mentioned fourth sub-region of interest image are diagonal images.

[0093] Step 3: Determine the set of sub-image thresholds corresponding to the above-mentioned set of sub-region of interest images according to the above-mentioned set of sub-region of interest images. Among them, the sub-region of interest images in the above-mentioned set of sub-region of interest images correspond one-to-one with the sub-image thresholds in the above-mentioned set of sub-image thresholds. In practice, the above-mentioned execution entity may determine the set of sub-image thresholds corresponding to the above-mentioned set of sub-region of interest images through an image threshold algorithm. Here, the above-mentioned execution entity may use the maximum between-class variance algorithm to determine the set of sub-image thresholds corresponding to the above-mentioned set of sub-region of interest images.

[0094] Step 4: Determine the set of sub-image diagonal thresholds according to the above-mentioned set of sub-image thresholds. In practice, the above-mentioned execution entity may determine the set of sub-image diagonal thresholds through the following steps:

[0095] The first sub-step: Determine the sub-image threshold corresponding to the above-mentioned first sub-region of interest image in the above-mentioned set of sub-image thresholds as the first sub-image threshold.

[0096] The second sub-step: Determine the sub-image threshold corresponding to the above-mentioned second sub-region of interest image in the above-mentioned set of sub-image thresholds as the second sub-image threshold.

[0097] The third sub-step: Determine the sub-image threshold corresponding to the above-mentioned third sub-region of interest image in the above-mentioned set of sub-image thresholds as the third sub-image threshold.

[0098] The fourth sub-step: Determine the sub-image threshold corresponding to the above-mentioned fourth sub-region of interest image in the above-mentioned set of sub-image thresholds as the fourth sub-image threshold.

[0099] The fifth sub-step is to determine the average value of the above-mentioned first sub-image threshold and the above-mentioned third sub-image threshold as the first sub-image diagonal threshold.

[0100] The sixth sub-step is to determine the average value of the above-mentioned second sub-image threshold and the above-mentioned fourth sub-image threshold as the second sub-image diagonal threshold.

[0101] The seventh sub-step is to determine the above-mentioned first sub-image diagonal threshold and the above-mentioned second sub-image diagonal threshold as the sub-image diagonal threshold set.

[0102] The fifth step is to perform binarization processing on each of the region-of-interest sub-image sets included in the above-mentioned region-of-interest sub-image set according to the above-mentioned sub-image diagonal threshold set to generate a binarized region image set. As an example, the binarized region image in the above-mentioned binarized region image set can be represented by the following formula:

[0103]

[0104]

[0105] where the above-mentioned img i (x, y) represents the pixel value after binarization processing. The above-mentioned i represents the region-of-interest sub-image in the region-of-interest sub-image set. The above-mentioned T 13 represents the first sub-image diagonal threshold. The above-mentioned T 24 represents the second sub-image diagonal threshold. The above-mentioned x represents the abscissa of the pixel point in the region-of-interest sub-image. The above-mentioned y represents the ordinate of the pixel point in the region-of-interest sub-image.

[0106] The sixth step is to generate a binarized region-of-interest image according to the above-mentioned binarized region image set. In practice, the above-mentioned execution entity can perform a combination process on the above-mentioned binarized region image set to generate a binarized region-of-interest image. Here, the combination method can be splicing.

[0107] The above first step to the sixth step and their related content are an inventive point of the embodiment of the present disclosure, which solves the second technical problem mentioned in the background art: "During the process of image processing, the uneven illumination of different parts of the captured image is not considered, and a preset constant threshold is used for binarization processing, resulting in poor clarity of the image after binarization processing, thereby leading to low accuracy of image feature matching, requiring multiple acquisitions of images, causing waste of image acquisition resources; at the same time, it causes inconvenience to users and poor user experience." The factors that lead to poor clarity of the image, thereby resulting in low accuracy of image feature matching, requiring multiple acquisitions of images, causing waste of image acquisition resources; at the same time, causing inconvenience to users and poor user experience are often as follows: During the process of image processing, the uneven illumination of different parts of the captured image is not considered, and a preset constant threshold is used for binarization processing, resulting in poor clarity of the image after binarization processing, thereby leading to low accuracy of image feature matching, requiring multiple acquisitions of images, causing waste of image acquisition resources; at the same time, it causes inconvenience to users and poor user experience. If the above factors are solved, the clarity of the image to be matched can be improved, the accuracy of image feature matching can be increased, the waste of image acquisition resources can be reduced, and the user experience can be improved. To achieve this effect, first, determine the center point coordinates of the above-mentioned region of interest image. Thus, the center point coordinates representing the center of the region of interest image can be obtained. Then, according to the above center point coordinates and the above region of interest image, determine the set of sub-region of interest images. Thus, the set of sub-region of interest images representing different parts of the region of interest image can be obtained. After that, according to the above set of sub-region of interest images, determine the set of sub-image thresholds corresponding to the above set of sub-region of interest images. Among them, the sub-region of interest images in the above set of sub-region of interest images correspond to the sub-image thresholds in the set of sub-image thresholds. Thus, the set of sub-image thresholds representing different illumination intensities of different parts of the region of interest image can be obtained. Secondly, according to the above set of sub-image thresholds, determine the set of sub-image diagonal thresholds. Thus, the set of sub-image diagonal thresholds can be obtained, so that the illumination intensity of different image parts can be adjusted. Then, according to the above set of sub-image diagonal thresholds, perform binarization processing on each sub-region of interest image included in the set of sub-region of interest images to generate a set of binarized region images. Thus, by using the threshold after illumination intensity equalization processing to perform binarization processing on the sub-region of interest images, a set of binarized region images can be obtained, which can be used to improve the clarity of the image. Finally, according to the above set of binarized region images, generate a binarized region of interest image. Thus, an image after different threshold equalization processing for different parts is obtained.Also, by segmenting the image of the region of interest, determining the mean of the diagonal image thresholds as the binarization threshold of the diagonal image, and performing binarization processing, the illumination intensity of different parts of the region of interest can be balanced. Thus, the clarity of the image after binarization processing can be improved, the accuracy of image feature matching can be enhanced, the number of times of image acquisition can be reduced, the waste of image acquisition resources can be decreased, and the user experience can be improved.

[0108] Step 108: Generate a palm vein matching result according to the feature vector of the region of interest and the preset feature vector set.

[0109] In some embodiments, according to the feature vector of the region of interest and the preset feature vector set, the above-mentioned execution entity can generate a palm vein matching result.

[0110] In some optional implementation manners of some embodiments, according to the feature vector of the region of interest and the preset feature vector set, the above-mentioned execution entity can generate a palm vein matching result through the following steps:

[0111] First step: For each preset feature vector in the above-mentioned preset feature vector set, determine the vector distance according to the above-mentioned preset feature vector and the feature vector of the region of interest. Wherein, the preset feature elements in the above-mentioned preset feature vector correspond one-to-one with the region of interest feature elements in the feature vector of the region of interest.

[0112] Second step: Determine the determined vector distances as a vector distance set.

[0113] Third step: Determine the vector distance with the smallest distance value in the above-mentioned vector distance set as the target vector distance.

[0114] Fourth step: In response to determining that the above-mentioned target vector distance is less than or equal to the preset distance threshold, determine the successful matching information indicating successful matching as the palm vein matching result of the above-mentioned user. Wherein, the above-mentioned preset distance threshold can be a preset distance threshold. For example, the above-mentioned preset distance threshold can be 10. The above-mentioned successful matching information can be information indicating that the palm vein feature information of the user matches the preset feature information stored in the preset intelligent door lock successfully. For example, the above-mentioned successful matching information can be "Palm information matches successfully. Welcome home, master".

[0115] Fifth step: In response to determining that the above-mentioned target vector distance is greater than the above-mentioned preset distance threshold, determine the failed matching information indicating failed matching as the palm vein matching result of the above-mentioned user. The above-mentioned failed matching information can be information indicating that the palm vein feature information of the user fails to match the preset feature information stored in the preset intelligent door lock. For example, the above-mentioned failed matching information can be "Palm information matches failed. Please re-match".

[0116] In some alternative implementations of some embodiments, based on the above-mentioned preset feature vector and the above-mentioned region of interest feature vector, the executing entity can determine the vector distance through the following steps:

[0117] First step, perform encoding processing on the above-mentioned preset feature vector to obtain a preset feature encoding vector. In practice, the executing entity can perform binary encoding processing on the above-mentioned preset feature vector to obtain the preset feature vector after binary encoding processing as the preset feature encoding vector.

[0118] Second step, perform encoding processing on the above-mentioned region of interest feature vector to obtain a region of interest feature encoding vector. Among them, the preset feature encoding elements in the above-mentioned preset feature encoding vector correspond one-to-one with the region of interest feature encoding elements in the above-mentioned region of interest feature encoding vector. In practice, the executing entity can perform binary encoding processing on the above-mentioned region of interest feature vector to obtain the region of interest feature vector after binary encoding processing as the region of interest feature encoding vector.

[0119] Third step, perform exclusive OR processing on the above-mentioned preset feature encoding vector and the above-mentioned region of interest feature encoding vector to obtain an exclusive OR numerical sequence. In practice, first, for each preset feature encoding element in the above-mentioned preset feature encoding vector, the executing entity can perform bitwise exclusive OR processing on the preset feature encoding element and the corresponding region of interest feature encoding element in the above-mentioned region of interest feature encoding vector to generate an exclusive OR value. Then, the generated exclusive OR values can be determined as the exclusive OR numerical sequence.

[0120] Fourth step, generate a vector distance according to the above-mentioned exclusive OR numerical sequence and a preset threshold. In practice, the executing entity can determine the number of exclusive OR values in the above-mentioned exclusive OR numerical sequence that are the same as the preset threshold as the vector distance.

[0121] Step 109, in response to determining that the palm vein matching result indicates a successful match, control the door lock component of the intelligent door lock to perform an unlocking operation.

[0122] In some embodiments, in response to determining that the above-mentioned palm vein matching result indicates a successful match, the executing entity can control the door lock component of the intelligent door lock to perform an unlocking operation. The door lock component can be a component for opening and closing the door lock. The unlocking operation can be an operation of unlocking.

[0123] Optionally, the above-mentioned execution entity may also control an associated sound playback device to play a re-matching operation prompt message in response to determining that the above-mentioned palm vein matching result indicates unsuccessful matching. Among them, the associated sound playback device may be a device for playing sound. For example, the above-mentioned associated sound playback device may be a power amplifier player.

[0124] The above-mentioned various embodiments of the present disclosure have the following beneficial effects: Through the intelligent door lock control method of some embodiments of the present disclosure, the clarity of the image to be matched can be improved, the accuracy of image feature matching can be enhanced, the waste of acquired image resources can be reduced, and the user experience can be improved. Specifically, the low clarity of the image leads to a low accuracy of image feature matching, requiring multiple acquisitions of images, resulting in a waste of acquired image resources; at the same time, it causes inconvenience to the user, and the reason for the poor user experience is that after the image is acquired, no image reconstruction operation is considered for the acquired image to improve the clarity of the image. The low clarity of the image leads to a low accuracy of image feature matching, requiring multiple acquisitions of images, resulting in a waste of acquired image resources; at the same time, it causes inconvenience to the user, and the user experience is poor. Based on this, in the intelligent door lock control method of some embodiments of the present disclosure, first, a palmar vein image of the user is acquired through a palmar vein image acquisition device. Thus, a palmar vein image can be obtained, which can then be used for matching and recognition. Then, the above-mentioned palmar vein image is input into the compressive sampling layer of a pre-trained image reconstruction model to obtain a compressive sensing measurement vector. Among them, the above-mentioned image reconstruction model further includes: an initial reconstructed image layer, a deep reconstructed image layer, and a global-local feature reconstructed image layer. Thus, a compressive sensing measurement vector after compression of the palmar vein image can be obtained. After that, the above-mentioned compressive sensing measurement vector is input into the above-mentioned initial reconstructed image layer to obtain an initial reconstructed image. Thus, through a pixel shuffling operation, an initial reconstructed image can be obtained. Subsequently, the above-mentioned initial reconstructed image is input into the above-mentioned deep reconstructed image layer to obtain a deep reconstructed image. Thus, a deep reconstructed image can be obtained. And through Fourier convolution operation, the clarity of the palmar vein image can be further improved. Secondly, the above-mentioned deep reconstructed image is input into the above-mentioned global-local feature reconstructed image layer to obtain a global-local feature reconstructed image as the palmar vein reconstructed image. Thus, a palmar vein reconstructed image can be obtained. Thereby, the palmar vein image can be reconstructed from two branches of global features and local features and fused, further improving the clarity of the constructed image. Then, a region of interest is extracted from the above-mentioned palmar vein reconstructed image to obtain a region of interest image. Thus, a region of interest image representing the target region can be obtained. Thereby, the region of the palmar vein image to be processed can be reduced, and the efficiency of processing the image can be improved. After that, features are extracted from the above-mentioned region of interest image to obtain a region of interest feature vector. Thus, a region of interest feature vector representing the palmar vein image can be obtained. Subsequently, a palmar vein matching result is generated according to the region of interest feature vector and a preset feature vector set. Thus, a palmar vein matching result can be obtained, which can then be used to determine whether the user is an authorized user. Finally, in response to determining that the above-mentioned palmar vein matching result indicates a successful match, the door lock component of the above-mentioned intelligent door lock is controlled to perform an unlocking operation. Thus, according to the matching result, the door lock can be controlled to perform an unlocking operation. Thereby, the user experience is improved.Also because after collecting the user's palm vein image, the palm vein image is subjected to depth reconstruction processing through Fourier convolution operation, which improves the clarity of the reconstructed image. Also because the palm vein image is reconstructed from two branches of global features and local features and fused, which further improves the clarity of the constructed image. Thereby improving the accuracy of image feature matching, reducing the number of image acquisitions, further reducing the waste of acquired image resources, and improving the user experience.

[0125] Further referring to Figure 2 , as an implementation of the methods shown in the above figures, the present disclosure provides some embodiments of a bridge intelligent door lock control device, and these device embodiments correspond to Figure 1 the method embodiments shown, and the device can be specifically applied to various intelligent door locks.

[0126] As Figure 2 shown, the intelligent door lock control device 200 of some embodiments includes: a collection unit 201, a first input unit 202, a second input unit 203, a third input unit 204, a fourth input unit 205, a region of interest extraction unit 206, a feature extraction unit 207, a generation unit 208, and a control unit 209. Among them, the collection unit 201 is configured to collect the user's palm vein image through a palm vein image collection device; the first input unit 202 is configured to input the above palm vein image into the compressive sampling layer of a pre-trained image reconstruction model to obtain a compressive sensing measurement vector, wherein the above image reconstruction model further includes: an initial reconstruction image layer, a depth reconstruction image layer, and a global-local feature reconstruction image layer; the second input unit 203 is configured to input the above compressive sensing measurement vector into the above initial reconstruction image layer to obtain an initial reconstruction image; the third input unit 204 is configured to input the above initial reconstruction image into the above depth reconstruction image layer to obtain a depth reconstruction image; the fourth input unit 205 is configured to input the above depth reconstruction image into the above global-local feature reconstruction image layer to obtain a global-local feature reconstruction image as a palm vein reconstruction image; the region of interest extraction unit 206 is configured to extract a region of interest from the above palm vein reconstruction image to obtain a region of interest image; the feature extraction unit 207 is configured to extract features from the above region of interest image to obtain a region of interest feature vector; the generation unit 208 is configured to generate a palm vein matching result according to the above region of interest feature vector and a preset feature vector set; the control unit 209 is configured to control the door lock component of the above intelligent door lock to perform an unlocking operation in response to determining that the above palm vein matching result indicates a successful match.

[0127] It can be understood that the units described in the intelligent door lock control device 200 and reference Figure 1corresponds to each step in the described method. Thus, the operations, features, and beneficial effects described above for the method also apply to the intelligent door lock control device 200 and the units included therein, and will not be elaborated herein.

[0128] Reference is now made to Figure 3 , which shows a schematic structural diagram of an intelligent door lock 300 suitable for use in implementing some embodiments of the present disclosure. The intelligent door lock in some embodiments of the present disclosure may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Tablet Computers), PMPs (Portable Multimedia Players), vehicle terminals (such as vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 3 The intelligent door lock shown is merely an example and should not impose any limitations on the functions and usage scope of the embodiments of the present disclosure.

[0129] As Figure 3 shown, the intelligent door lock 300 may include a processing device 301 (such as a central processing unit, a graphics processing unit, etc.), which may perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 into a random access memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the intelligent door lock 300 are also stored. The processing device 301, the ROM 302, and the RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0130] Generally, the following devices may be connected to the I / O interface 305: an input device 306 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 307 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; and a communication device 309. The communication device 309 may allow the intelligent door lock 300 to communicate with other devices wirelessly or wiredly to exchange data. Although Figure 3 shows an intelligent door lock 300 having various devices, it should be understood that it is not required to implement or include all the shown devices. Instead, more or fewer devices may be implemented or included. Figure 3 Each block shown in

[0131] In particular, according to some embodiments of the present disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of the present disclosure include a computer program product that includes a computer program carried on a computer-readable medium, and the computer program includes program code for performing the methods shown in the flowcharts. In such some embodiments, the computer program can be downloaded and installed from a network via a communication device 309, or installed from a storage device 308, or installed from a ROM 302. When the computer program is executed by a processing device 301, the above functions defined in the methods of some embodiments of the present disclosure are performed.

[0132] It should be noted that the computer-readable medium described in some embodiments of the present disclosure can be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can 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 of a computer-readable storage medium can include, but are not limited to: an electrical connection having one or more wires, a portable computer 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 disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In some embodiments of the present disclosure, the computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In some embodiments of the present disclosure, a computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, and the computer-readable signal medium 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 computer-readable medium can be transmitted by any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0133] In some embodiments, the client and the server can communicate using any currently known or future-developed network protocol such as HTTP (Hyper Text Transfer Protocol), and can be interconnected with digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include local area networks ("LANs"), wide area networks ("WANs"), the Internet (e.g., the Internet), and end-to-end networks (e.g., ad hoc end-to-end networks), as well as any currently known or future-developed networks.

[0134] The above computer-readable medium can be included in the above intelligent door lock; or it can exist separately and not be assembled into the intelligent door lock. The above computer-readable medium carries one or more programs. When the above one or more programs are executed by the intelligent door lock, the intelligent door lock is caused to: collect a palm vein image of a user through a palm vein image acquisition device; input the above palm vein image into a compressive sampling layer of a pre-trained image reconstruction model to obtain a compressive sensing measurement vector, wherein the above image reconstruction model further includes: an initial reconstructed image layer, a deep reconstructed image layer, and a global-local feature reconstructed image layer; input the above compressive sensing measurement vector into the above initial reconstructed image layer to obtain an initial reconstructed image; input the above initial reconstructed image into the above deep reconstructed image layer to obtain a deep reconstructed image; input the above deep reconstructed image into the above global-local feature reconstructed image layer to obtain a global-local feature reconstructed image as a palm vein reconstructed image; extract a region of interest from the above palm vein reconstructed image to obtain a region-of-interest image; extract features from the above region-of-interest image to obtain a region-of-interest feature vector; generate a palm vein matching result according to the region-of-interest feature vector and a preset feature vector set; and in response to determining that the above palm vein matching result indicates a successful match, control a door lock component of the above intelligent door lock to perform an unlocking operation.

[0135] Computer program code for performing the operations of some embodiments of the present disclosure can be written in one or more programming languages or combinations thereof. The above programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., by using an Internet service provider to connect through the Internet).

[0136] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a portion of code that contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions noted in the blocks may occur in a different order than noted in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, or they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0137] The functions described above can be performed, at least in part, by one or more hardware logic components. By way of example, and without limitation, the types of hardware logic components that may be used include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems on a chip (SOCs), complex programmable logic devices (CPLDs), and the like.

[0138] The above description is only some preferred embodiments of the present disclosure and an explanation of the technical principles applied. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by the specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the technical solutions formed by replacing the above features with (but not limited to) technical features having similar functions disclosed in the embodiments of the present disclosure.

Claims

1. An intelligent door lock control method, comprising: Collecting a palm vein image of a user through a palm vein image acquisition device; Inputting the palm vein image into a compressive sampling layer of a pre-trained image reconstruction model to obtain a compressive sensing measurement vector, wherein the image reconstruction model further includes: an initial reconstructed image layer, a deep reconstructed image layer, and a global-local feature reconstructed image layer; Inputting the compressive sensing measurement vector into the initial reconstructed image layer to obtain an initial reconstructed image; Inputting the initial reconstructed image into the deep reconstructed image layer to obtain a deep reconstructed image; Inputting the deep reconstructed image into the global-local feature reconstructed image layer to obtain a global-local feature reconstructed image as a palm vein reconstructed image; Performing region of interest extraction on the palm vein reconstructed image to obtain a region of interest image; Performing feature extraction on the region of interest image to obtain a region of interest feature vector; Generating a palm vein matching result according to the region of interest feature vector and a preset feature vector set; In response to determining that the palm vein matching result indicates successful matching, controlling a door lock component of the intelligent door lock to perform an unlocking operation.

2. The method according to claim 1, wherein, Before performing the feature extraction on the region of interest image to obtain a region of interest feature vector, the method further includes: Performing binarization processing on the region of interest image to obtain a binarized region of interest image as a binarized region of interest image; Performing contrast stretching processing on the binarized region of interest image to obtain a contrast-stretched binarized region of interest image as a contrast-stretched processed image; Performing image enhancement processing on the contrast-stretched processed image to generate an enhanced region of interest image; Determining the enhanced region of interest image as the region of interest image to update the region of interest image.

3. The method according to claim 1, wherein, The method further includes: In response to determining that the palm vein matching result indicates unsuccessful matching, controlling an associated sound playback device to play a re-matching operation prompt message.

4. The method according to claim 1, wherein The generating a palm vein matching result according to the region of interest feature vector and a preset feature vector set includes: For each preset feature vector in the preset feature vector set, determining a vector distance according to the preset feature vector and the region of interest feature vector, wherein preset feature elements in the preset feature vector correspond to region of interest feature elements in the region of interest feature vector; Determining the determined respective vector distances as a vector distance set; Determining the vector distance with the smallest distance value in the vector distance set as a target vector distance; In response to determining that the target vector distance is less than or equal to a preset distance threshold, determining a successful matching message indicating successful matching as the palm vein matching result of the user; In response to determining that the target vector distance is greater than the preset distance threshold, determining a failed matching message indicating failed matching as the palm vein matching result of the user.

5. The method according to claim 1, wherein, The performing region of interest extraction on the palm vein reconstructed image to obtain a region of interest image includes: Generate palm key point information according to the reconstructed palm vein image and the pre-trained palm key point detection model, where the palm key point information includes: first valley bottom key point information, second valley bottom key point information, and wrist center key point information. The first valley bottom key point information is the coordinate information of the valley bottom between the index finger and the middle finger, and the second valley bottom key point information is the coordinate information of the valley bottom between the middle finger and the ring finger; Determine the image rotation angle according to the first valley bottom key point information and the second valley bottom key point information; Perform a rotation process on the preprocessed palm vein image according to the rotation angle to generate a rotated image; Determine the first and second valley bottom midpoint information according to the first valley bottom key point information and the second valley bottom key point information; Determine the palm center coordinate information according to the first and second valley bottom midpoint information and the wrist center key point information; Determine the region of interest image of the user according to the palm center coordinate information and the rotated image; 6. The method according to claim 5, wherein, The determining the image rotation angle according to the first valley bottom key point information and the second valley bottom key point information includes: Determine the abscissa difference information by taking the difference between the abscissa included in the first valley bottom key point information and the abscissa included in the second valley bottom key point information; Determine the ordinate difference information by taking the difference between the ordinate included in the first valley bottom key point information and the ordinate included in the second valley bottom key point information; Determine the coordinate difference information by taking the ratio of the abscissa difference information to the ordinate difference information; Determine the image rotation angle as the angle corresponding to the arctangent of the coordinate difference information.

7. The method according to claim 4, wherein, The determining the vector distance according to the preset feature vector and the region of interest feature vector includes: Perform an encoding process on the preset feature vector to obtain a preset feature encoding vector; Perform an encoding process on the region of interest feature vector to obtain a region of interest feature encoding vector, where the preset feature encoding elements in the preset feature encoding vector correspond to the region of interest feature encoding elements in the region of interest feature encoding vector; Perform an exclusive OR process on the preset feature encoding vector and the region of interest feature encoding vector to obtain an exclusive OR numerical sequence; Generate a vector distance according to the exclusive OR numerical sequence and a preset threshold; 8. The method according to claim 2, wherein, The performing a binarization process on the region of interest image to obtain a binarized region of interest image as the binarized region of interest image includes: Determine the center point coordinates of the region of interest image; Determine a set of region of interest sub-image according to the center point coordinates and the region of interest image; Determine a set of sub-image thresholds corresponding to the set of region of interest sub-images according to the set of region of interest sub-images, where the region of interest sub-images in the set of region of interest sub-images correspond to the sub-image thresholds in the set of sub-image thresholds; Determine a set of sub-image diagonal thresholds according to the set of sub-image thresholds; According to the set of diagonal thresholds of the sub-images, perform binarization processing on each sub-region of interest image included in the set of sub-regions of interest images to generate a set of binarized region images; Generate a binarized region of interest image according to the set of binarized region images.

9. An intelligent door lock, comprising: One or more processors; A door lock assembly; A storage device on which one or more programs are stored, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1-8.

10. A computer-readable medium having a computer program stored thereon, wherein, When the computer program is executed by a processor, the method according to any one of claims 1-8 is implemented.

Images