Low-computing-power face encryption transmission method and system and related equipment

Abstract

The invention provides a low-computing-power face encryption transmission method and system and related equipment, and the method comprises the steps: collecting a face image, and obtaining face image data; establishing a lightweight face detection model based on a dbface algorithm, and detecting a face part in the face image data through the lightweight face detection model to obtain local face coordinate data; encrypting a face part in the face image data according to the local face coordinate data to obtain encrypted face image data; and decrypting and transmitting the encrypted face image data. Compared with the prior art, through fusion of the lightweight face detection model and the encryption algorithm, instant face detection and encryption requirements can be realized under the condition of low computing power resources, and meanwhile, due to the fact that only the face part in the face image is subjected to in-situ local encryption, the storage space consumption is reduced, and the encryption efficiency is improved. And the requirements of carrying out instant encryption protection and decryption recovery on private faces in real-time image streams and video streams are met.

Description

Technical Field

[0001] The present invention is applicable to the field of face encryption technology, and in particular relates to a low-computing-power face encryption transmission method, system and related equipment. Background Art

[0002] With the improvement of computing power and the continuous advancement of deep learning technology, face recognition technology has been widely used in many fields such as security, finance, and medical care. However, as face recognition technology becomes more popular, how to ensure the security and privacy of user face data has become an urgent problem to be solved.

[0003] The transmission of face data mainly includes face detection, face encryption, data transmission and face decryption. In the existing related technologies, face detection is often based on deep learning. Face detection algorithms are often used to determine the position of the face, extract information such as face size and posture, and lay the foundation for further advanced applications such as face recognition, emotion analysis, and identity authentication. Its implementation process usually involves multiple steps such as image preprocessing, feature extraction, candidate area generation, and classification judgment. Face detection based on deep learning can be divided into the following four categories:

[0004] 1. The cascaded convolutional neural network (CNN) model. The core idea of ​​this type of model is to gradually screen and refine the detection results from coarse to fine. It is usually composed of multiple sub-networks, each of which is responsible for processing tasks of different difficulty levels. The front sub-network quickly excludes a large number of non-face areas, while the back sub-network performs a more detailed analysis of the remaining candidate areas.

[0005] 2. Regional Convolutional Neural Network (R-CNN) Model,The first stage of this scheme is the multi-task region proposal network (RPN), which generates a large number of candidate regions from the input image and extracts the features of each region. The second stage is R-CNN, which verifies whether the candidate region is a valid face.

[0006] 3. Single-stage detection (SSD) model. This scheme first uses CNN to generate multiple feature maps of different scales. Each feature map corresponds to a specific spatial resolution of the original image, and a set of default boxes are pre-set at each position on each feature map. Then, the predicted bounding box and classification score are output, and whether it is a face area is determined based on the score.

[0007] Fourth, the core idea of ​​the model based on feature pyramid network is to utilize the inherent multi-scale features of a single convolutional neural network, aggregate high-level semantic feature maps of different scales as context clues, and then enhance low-level feature maps through hierarchical aggregation with marginal additional computational cost, so as to capture key faces.

[0008] Among the existing mainstream cryptographic algorithms, there are three main categories: 1. Symmetric encryption. This type of algorithm uses the same key for encryption and decryption, and is often encrypted through a series of reversible transformations (such as permutation, XOR), and is usually used to encrypt large amounts of data. Typical algorithms include Data Encryption Standard (DES) and Advanced Encryption Standard (AES). 2. Asymmetric encryption. This type of algorithm uses a pair of keys - public key and private key, where the public key can be shared publicly, while the private key must be kept confidential. It is generally based on more complex mathematical principles (such as large integer factorization, discrete logarithm problems on elliptic curves), and is usually used to encrypt small amounts of data, key exchange and digital signatures. Typical algorithms include RSA and elliptic curve cryptography (ECC). 3. Hash function. This type of algorithm uses a hash function to generate a fixed-length output (called a digest or hash value), which is usually irreversible, and is therefore used for data integrity checking and password storage. Typical algorithms include the MD5 message digest algorithm and the secure hash algorithm (SHA).

[0009] For existing face detection-encryption systems, face eigenvalues ​​are usually used as an intermediary. Specifically, the system first uses a face detection algorithm to locate the face area, then runs an eigenvalue algorithm to calculate the corresponding eigenvalue for the face area, and finally encrypts the eigenvalue with the help of a cryptographic algorithm.

[0010] However, the existing face detection-encryption system has the following three disadvantages: 1. The face detection algorithm is computationally expensive: The core performance of the mainstream face detection model is detection accuracy. The model is designed based on a highly complex model. Not only does it have a large number of parameters and a long training time, but it also relies on the support of high-performance computing power (such as GPU) to ensure the detection speed. However, in actual application scenarios, it is often necessary to execute the detection algorithm on low-computing devices at the edge. Achieving efficient face detection under low computing resources is an important application-oriented issue. 2. Standard encryption method computational redundancy: As the cornerstone of information security, encryption algorithms ensure the confidentiality and integrity of data during network transmission and storage. However, existing encryption algorithms often have a high computational cost and are not specifically considered and designed for image and face data. If used to encrypt complete images, it will bring additional computing power consumption. For time-sensitive scenarios such as real-time image streams and video streams, it is difficult to protect them by directly deploying traditional encryption algorithms. 3. Existing systems are limited to protecting features: The current mainstream dense face detection system essentially only encrypts the facial feature values, and cannot achieve in-situ encryption of the facial area, and cannot meet the needs of instant encryption protection and decryption recovery of private facial images in real-time image streams and video streams. In general, using the current mainstream face detection model and reliable traditional encryption algorithms, the existing dense face detection solutions cannot meet the performance requirements of instant face detection, encrypted transmission, and decryption recovery in real-time images or video streams in low computing power scenarios.

[0011] Therefore, there is an urgent need for a new low-computing power face encryption transmission method, system and related equipment to solve the above technical problems. Summary of the invention

[0012] The present invention provides a low-computing-power face encryption transmission method, system and related equipment, aiming to achieve efficient face data encryption transmission in low-computing-power scenarios.

[0013] In a first aspect, the present invention provides a low-computing-power face encryption transmission method, the face encryption method comprising the following steps:

[0014] S1. Collecting facial images to obtain facial image data;

[0015] S2. Establishing a lightweight face detection model based on the dbface algorithm, detecting the face part in the face image data with a preset dimension through the lightweight face detection model to obtain local face coordinate data in the face image data;

[0016] S3, locally encrypting the face portion in the face image data according to the local face coordinate data to obtain encrypted face image data;

[0017] S4. Send the encrypted facial image data to the target end for decryption.

[0018] Preferably, step S2 includes the following sub-steps:

[0019] S21, normalizing the facial image data to obtain pre-processed facial image data;

[0020] S22, performing face detection of preset dimensions on the face part of the pre-processed face image data to obtain the local face coordinate data; wherein the local face coordinate data includes face frame center point data, face frame size data and face key point data.

[0021] Preferably, step S3 includes the following sub-steps:

[0022] S31, serializing the face part in the face image data according to the local face coordinate data to obtain plaintext pixel data of the face image;

[0023] S32, performing sequence encryption on the plaintext pixel data of the face image based on an encryption algorithm to obtain ciphertext pixel data of the face image;

[0024] S33. Replace the face part in the face image data according to the face image ciphertext pixel data to obtain the encrypted face image data.

[0025] Preferably, the encryption algorithm is based on the AES algorithm and the SM4 algorithm.

[0026] In a second aspect, the present invention further provides a low-computing-power face encryption transmission system, comprising:

[0027] A face acquisition module is used to acquire face images and obtain face image data;

[0028] A face detection module is used to establish a lightweight face detection model based on the dbface algorithm, and detect the face part in the face image data with a preset dimension through the lightweight face detection model to obtain local face coordinate data in the face image data;

[0029] A local encryption module, used for locally encrypting the face part in the face image data according to the local face coordinate data to obtain encrypted face image data;

[0030] The decryption module is used to send the encrypted face image data to the target end for decryption.

[0031] Preferably, the face detection module includes a preprocessing unit and a detection unit;

[0032] The preprocessing unit is used to perform normalization processing on the face image data to obtain preprocessed face image data;

[0033] The detection unit is used to perform face detection of preset dimensions on the face part in the pre-processed face image data to obtain the local face coordinate data; wherein the local face coordinate data includes face frame center point data, face frame size data and face key point data.

[0034] Preferably, the local encryption module includes a serialization unit, a sequence encryption unit and a replacement unit;

[0035] The serialization unit is used to serialize the face part in the face image data according to the local face coordinate data to obtain plain text pixel data of the face image;

[0036] The sequence encryption unit is used to perform sequence encryption on the plaintext pixel data of the face image based on an encryption algorithm to obtain ciphertext pixel data of the face image;

[0037] The replacement unit is used to replace the face part in the face image data according to the face image ciphertext pixel data to obtain the encrypted face image data.

[0038] Preferably, the encryption algorithm is based on the AES algorithm and the SM4 algorithm.

[0039] In the third aspect, the present invention also provides a computer device, comprising: a memory, a processor, and a low-computing power face encryption transmission program stored on the memory and runable on the processor, wherein when the processor executes the low-computing power face encryption transmission program, it implements the steps in the low-computing power face encryption transmission method as described in any one of the above embodiments.

[0040] In a fourth aspect, the present invention also provides a computer-readable storage medium, on which a low-computing power face encryption transmission program is stored. When the low-computing power face encryption transmission program is executed by a processor, the steps in the low-computing power face encryption transmission method described in any one of the above embodiments are implemented.

[0041] Compared with the prior art, the present invention acquires face image data by collecting face images; establishes a lightweight face detection model based on the dbface algorithm, detects the face part in the face image data through the lightweight face detection model, and acquires local face coordinate data; encrypts the face part in the face image data according to the local face coordinate data to acquire encrypted face image data; and decrypts and transmits the encrypted face image data. By integrating the lightweight face detection model and the encryption algorithm, the present invention can realize instant face detection and encryption requirements under low computing power resources. At the same time, since only the face part in the face image is locally encrypted in situ, the storage space consumption is reduced, and the requirements for instant encryption protection and decryption recovery of private faces in real-time image streams and video streams are met. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The present invention will be described in detail below in conjunction with the accompanying drawings. The above and other aspects of the present invention will become clearer and easier to understand through the detailed description made in conjunction with the following drawings. In the accompanying drawings:

[0043] Figure 1 It is a flowchart of a low-computing-power face encryption transmission method provided by an embodiment of the present invention;

[0044] Figure 2 This is a performance comparison diagram of the lightweight face detection model of the low-computing-power face encryption transmission method provided by an embodiment of the present invention and the face detection model of the related art;

[0045] Figure 3 This is a performance comparison chart of the encryption algorithm of the low-computing-power face encryption transmission method provided by an embodiment of the present invention and the encryption algorithm of the related art;

[0046] Figure 4 It is a structural schematic diagram of a low-computing-power face encryption transmission system provided by an embodiment of the present invention;

[0047] Figure 5It is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0049] Embodiment 1

[0050] Please refer to Figure 1 The present invention provides a low-computing-power face encryption transmission method, the face encryption method comprising the following steps:

[0051] S1. Collecting facial images to obtain facial image data.

[0052] In the embodiment of the present invention, the facial image data can be collected and acquired through a visual sensor.

[0053] S2. Establishing a lightweight face detection model based on the dbface algorithm, detecting the face part in the face image data with a preset dimension through the lightweight face detection model to obtain local face coordinate data in the face image data;

[0054] In the embodiment of the present invention, step S2 includes the following sub-steps:

[0055] S21, normalizing the facial image data to obtain pre-processed facial image data;

[0056] S22, perform face detection of preset dimensions on the face part in the pre-processed face image data to obtain the local face coordinate data; wherein the local face coordinate data is the data related to the face part in the face image data, and the local face coordinate data includes face frame center point data, face frame size data and face key point data. Specifically, the face frame center point data includes the coordinates of the center position of the face in the face image data, the face frame size data includes multiple coordinate points after a certain offset relative to the face frame center point, and the multiple coordinate points can frame the face part in the face image data, and the face key point data includes the feature points such as eyes, nose and mouth of the face in the face data image.

[0057] The lightweight face detection model of the present invention is based on the dbface model. The model structure adopted by the mainstream solutions in the industry is complex, with many parameters, large consumption of computing resources, and reliance on GPU acceleration. The present invention focuses on the application of face detection on end-side devices such as embedded and mobile phones with limited computing resources, which requires speed and performance optimization. It adopts a lightweight and efficient face detection model, which can achieve real-time face detection in seconds on a single CPU. The lightweight detection model can significantly reduce the computational complexity and model size while maintaining a high detection accuracy.

[0058] In order to verify the advantages of lightweight algorithms over non-lightweight algorithms in project scenarios, this paper conducts comparative experiments on the commonly used image dataset FDDB. For detailed comprehensive performance comparison, see Figure 2 When computing power is limited to a single-core CPU without a GPU, the single-image detection time of the mainstream non-lightweight representative solution DSFD is 20 to 200 times longer than that of the lightweight algorithm; at the same time, this cost only brings an additional 3% performance improvement compared to the lightweight algorithms (such as the facebox algorithm, lffd algorithm, mtcnn algorithm, retinaface algorithm, centerface algorithm, libface algorithm, yoloface algorithm and blazeface algorithm in the figure), and the cost performance is low.

[0059] The present invention selects dbface with the best comprehensive performance as a lightweight face detection model, and additionally provides 9 other spare lightweight algorithms for secondary testing and development. Non-lightweight face detection models represented by DSFD often use Anchor Based face detection methods and deep models with large parameters as the backbone network, which causes such methods to be difficult to train not only because of the large number of parameters (greater than the order of 100M), but also have significantly slower face detection speed in scenarios with low computing power resources. In contrast, dbface used in the present invention is based on a modern convolutional neural network (CNN) architecture, uses an Anchor Free detection method, and integrates advanced face detection models such as CenterNet and high-performance face recognition algorithms (such as ArcFace), which not only trains faster, but also can maintain considerable face detection speed and accuracy in scenarios with low computing power resources. Among them, the CenterNet network structure, as the core of the face detection algorithm, replaces the Anchor Box in the Anchor Based face detection algorithm by taking the center point as the reference point, and performs prediction in three branches, avoiding the high sampling cost and complex parameter adjustment process, and can work effectively even in complex environments; ArcFace, as a loss function, introduces the angular distance in the cosine space during the training process, enhancing the model's robustness to different facial expressions, postures, and lighting changes. On this basis, dbface further optimized its small model architecture, and its backbone network was selected as MobileNetV3, which consists of only a few hundred neural network layer structures, and the overall model size is only 7MB, which can run on devices with limited resources.

[0060] S3, locally encrypting the face portion in the face image data according to the local face coordinate data to obtain encrypted face image data;

[0061] In the embodiment of the present invention, step S3 includes the following sub-steps:

[0062] S31, serializing the face part in the face image data according to the local face coordinate data to obtain plaintext pixel data of the face image;

[0063] S32, performing sequence encryption on the plaintext pixel data of the face image based on an encryption algorithm to obtain ciphertext pixel data of the face image;

[0064] S33. Replace the face part in the face image data according to the face image ciphertext pixel data to obtain the encrypted face image data.

[0065] In an embodiment of the present invention, the encryption algorithm is based on the AES algorithm and the SM4 algorithm, and is also combined with the application of an image obfuscation algorithm. By performing local encryption, only the sensitive area (face part) in the image is encrypted, and the visibility of other non-sensitive parts is maintained. Since the face area often only occupies a small part of the image, the computational complexity of local encryption is obviously lower. In the scenario of encrypting a large number of face images, global encryption and decryption operations will consume more computing and memory resources, affecting user experience and service efficiency. In contrast, local encryption not only protects key information, but also retains some readability of the image, reduces redundant operations, and has application advantages in end-side trusted face recognition scenarios with limited computing power.

[0066] The present invention carried out comparative experiments on the FDDB dataset. The specific performance comparison results are as follows: Figure 3 As shown. For the SM4 algorithm, global encryption takes nearly 5 times more time to encrypt the complete FDDB data set; for the AES algorithm, global encryption takes 1.6 times as long as local in-situ encryption. In addition, the image obfuscation algorithms involved in the comparison show relatively poor encryption and decryption performance. The main reason is that as an emerging encryption algorithm, the industry's research and application of it is relatively insufficient, and there is a lack of high-quality implementation of related algorithms. There is only a python code implementation with low computational efficiency; the research and application of traditional global encryption algorithms are relatively mature. Although python library functions are used, their underlying implementation is a C language library with higher computational efficiency, which results in their significant advantages over image obfuscation algorithms.

[0067] S4. Send the encrypted facial image data to the target end for decryption.

[0068] In the embodiment of the present invention, since a lightweight face detection model based on dbface is used, which is composed of only a few hundred neural network layer structures, the overall model size is only 7MB, and it can be run on devices with limited resources. At the same time, only the face part of the face image data is encrypted through local encryption, so that the real-time face detection and encryption requirements can be achieved under the condition of low computing power resources. After the encrypted face image data is stored locally, it is transmitted through the cloud for local in-situ decryption, thereby achieving the real-time encryption protection and decryption recovery requirements of private face images in real-time image streams and video streams.

[0069] Compared with the prior art, the present invention acquires face image data by collecting face images; establishes a lightweight face detection model based on the dbface algorithm, detects the face part in the face image data through the lightweight face detection model, and acquires local face coordinate data; encrypts the face part in the face image data according to the local face coordinate data to acquire encrypted face image data; and decrypts and transmits the encrypted face image data. By integrating the lightweight face detection model and the encryption algorithm, the present invention can realize instant face detection and encryption requirements under low computing power resources. At the same time, since only the face part in the face image is locally encrypted in situ, the storage space consumption is reduced, and the requirements for instant encryption protection and decryption recovery of private faces in real-time image streams and video streams are met.

[0070] Embodiment 2

[0071] The embodiment of the present invention also provides a low-computing-power face encryption transmission system, please refer to Figure 4 , Figure 4 : is a structural diagram of a low-computing-power face encryption transmission system 200 provided in an embodiment of the present invention, which includes:

[0072] 201. A face acquisition module, used for acquiring face images to obtain face image data;

[0073] 202. A face detection module, configured to establish a lightweight face detection model based on a dbface algorithm, and detect a face part in the face image data in a preset dimension by using the lightweight face detection model to obtain local face coordinate data in the face image data;

[0074] 203. A local encryption module, configured to locally encrypt the face portion of the face image data according to the local face coordinate data to obtain encrypted face image data;

[0075] 204. A decryption module, used to send the encrypted facial image data to a target end for decryption.

[0076] In the embodiment of the present invention, the face detection module 202 includes a preprocessing unit 2021 and a detection unit 2022;

[0077] The preprocessing unit 2021 is used to perform normalization processing on the face image data to obtain preprocessed face image data;

[0078] The detection unit 2022 is used to perform face detection of preset dimensions on the face part in the pre-processed face image data to obtain the local face coordinate data; wherein the local face coordinate data includes face frame center point data, face frame size data and face key point data.

[0079] In the embodiment of the present invention, the local encryption module 203 includes a serialization unit 2031, a sequence encryption unit 2032 and a replacement unit 2033;

[0080] The serialization unit 2031 is used to serialize the face part in the face image data according to the local face coordinate data to obtain plain text pixel data of the face image;

[0081] The sequence encryption unit 2032 is used to perform sequence encryption on the plaintext pixel data of the face image based on an encryption algorithm to obtain ciphertext pixel data of the face image;

[0082] The replacement unit 2033 is used to replace the face part in the face image data according to the face image ciphertext pixel data to obtain the encrypted face image data.

[0083] In the embodiment of the present invention, the encryption algorithm is based on the AES algorithm and the SM4 algorithm.

[0084] The low-computing-power face encryption transmission system 200 can implement the steps in the low-computing-power face encryption transmission method in the above-mentioned embodiment, and can achieve the same technical effects. Please refer to the description in the above-mentioned embodiment and will not be repeated here.

[0085] Embodiment 3

[0086] The embodiment of the present invention also provides a computer device, please refer to Figure 5 , Figure 5 It is a structural diagram of a computer device provided in an embodiment of the present invention, wherein the computer device 300 comprises: a memory 302, a processor 301, and a low-computing-power face encryption transmission program stored in the memory 302 and executable on the processor 301.

[0087] The processor 301 calls the low-computing-power face encryption transmission program stored in the memory 302 to execute the steps in the low-computing-power face encryption transmission method provided in the embodiment of the present invention. Figure 1 , specifically including the following steps:

[0088] S1. Collecting facial images to obtain facial image data.

[0089] In the embodiment of the present invention, the facial image data can be collected and acquired through a visual sensor.

[0090] S2. Establishing a lightweight face detection model based on the dbface algorithm, detecting the face part in the face image data with a preset dimension through the lightweight face detection model to obtain local face coordinate data in the face image data;

[0091] In the embodiment of the present invention, step S2 includes the following sub-steps:

[0092] S21, normalizing the facial image data to obtain pre-processed facial image data;

[0093] S22, perform face detection of preset dimensions on the face part in the pre-processed face image data to obtain the local face coordinate data; wherein the local face coordinate data includes face frame center point data, face frame size data and face key point data. Specifically, the face frame center point data includes the coordinates of the center position of the face in the face image data, the face frame size data includes multiple coordinate points after a certain offset relative to the face frame center point, and the multiple coordinate points can frame the face part in the face image data, and the face key point data includes feature points such as eyes, nose and mouth of the face in the face data image.

[0094] The lightweight face detection model of the present invention is based on the dbface model. The model structure adopted by the mainstream solutions in the industry is complex, with many parameters, large consumption of computing resources, and reliance on GPU acceleration. The present invention focuses on the application of face detection on end-side devices such as embedded and mobile phones with limited computing resources, which requires speed and performance optimization. It adopts a lightweight and efficient face detection model, which can achieve real-time face detection in seconds on a single CPU. The lightweight detection model can significantly reduce the computational complexity and model size while maintaining a high detection accuracy.

[0095] S3, locally encrypting the face portion in the face image data according to the local face coordinate data to obtain encrypted face image data;

[0096] In the embodiment of the present invention, step S3 includes the following sub-steps:

[0097] S31, serializing the face part in the face image data according to the local face coordinate data to obtain plaintext pixel data of the face image;

[0098] S32, encrypting the plaintext pixel data of the face image in sequence based on an encryption algorithm to obtain ciphertext pixel data of the face image;

[0099] S33. Replace the face part in the face image data according to the face image ciphertext pixel data to obtain the encrypted face image data.

[0100] In an embodiment of the present invention, the encryption algorithm is based on the AES algorithm and the SM4 algorithm, and is also combined with the application of an image obfuscation algorithm. By performing local encryption, only the sensitive area (face part) in the image is encrypted, and the visibility of other non-sensitive parts is maintained. Since the face area often only occupies a small part of the image, the computational complexity of local encryption is obviously lower. In the scenario of encrypting a large number of face images, global encryption and decryption operations will consume more computing and memory resources, affecting user experience and service efficiency. In contrast, local encryption not only protects key information, but also retains some readability of the image, reduces redundant operations, and has application advantages in end-side trusted face recognition scenarios with limited computing power.

[0101] S4. Send the encrypted facial image data to the target end for decryption.

[0102] In the embodiment of the present invention, since a lightweight face detection model based on dbface is used, which is composed of only a few hundred neural network layer structures, the overall model size is only 7MB, and it can be run on devices with limited resources. At the same time, only the face part of the face image data is encrypted through local encryption, so that the real-time face detection and encryption requirements can be achieved under the condition of low computing power resources. After the encrypted face image data is stored locally, it is transmitted through the cloud for local in-situ decryption, thereby achieving the real-time encryption protection and decryption recovery requirements of private face images in real-time image streams and video streams.

[0103] The computer device 300 provided in an embodiment of the present invention can implement the steps in the low-computing power face encryption transmission method in the above embodiment, and can achieve the same technical effect. Refer to the description in the above embodiment and will not be repeated here.

[0104] Embodiment 4

[0105] An embodiment of the present invention also provides a computer-readable storage medium, which stores a low-computing power face encryption transmission program. When the low-computing power face encryption transmission program is executed by a processor, it implements the various processes and steps in the low-computing power face encryption transmission method provided by an embodiment of the present invention, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0106] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM).

[0107] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0108] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0109] The embodiments of the present invention are described above in conjunction with the accompanying drawings. What is disclosed is only the preferred embodiment of the present invention. However, the present invention is not limited to the above-mentioned specific implementation manner. The above-mentioned specific implementation manner is only illustrative rather than restrictive. Under the enlightenment of the present invention, ordinary technicians in this field can also make many forms and equivalent changes without departing from the scope of protection of the purpose of the present invention and the claims, all of which are within the protection of the present invention.

Claims

1. A low-computing-power face encryption transmission method, characterized in that: The face encryption method comprises the following steps: S1. Collecting facial images to obtain facial image data; S2. Establishing a lightweight face detection model based on the dbface algorithm, detecting the face part in the face image data with a preset dimension through the lightweight face detection model to obtain local face coordinate data in the face image data; S3, locally encrypting the face portion in the face image data according to the local face coordinate data to obtain encrypted face image data; S4. Send the encrypted facial image data to the target end for decryption.

2. The low-computing-power face encryption transmission method according to claim 1, characterized in that: Step S2 includes the following sub-steps: S21, normalizing the facial image data to obtain pre-processed facial image data; S22, performing face detection of preset dimensions on the face part of the pre-processed face image data to obtain the local face coordinate data; wherein the local face coordinate data includes face frame center point data, face frame size data and face key point data.

3. The low-computing-power face encryption transmission method according to claim 1, characterized in that: Step S3 includes the following sub-steps: S31, serializing the face part in the face image data according to the local face coordinate data to obtain plaintext pixel data of the face image; S32, performing sequence encryption on the plaintext pixel data of the face image based on an encryption algorithm to obtain ciphertext pixel data of the face image; S33. Replace the face part in the face image data according to the face image ciphertext pixel data to obtain the encrypted face image data.

4. The low-computing-power face encryption transmission method as claimed in claim 3, characterized in that: The encryption algorithm is based on the AES algorithm and the SM4 algorithm.

5. A low-computing-power face encryption transmission system, characterized in that: include: A face acquisition module is used to acquire face images and obtain face image data; A face detection module is used to establish a lightweight face detection model based on the dbface algorithm, and detect the face part in the face image data with a preset dimension through the lightweight face detection model to obtain local face coordinate data in the face image data; A local encryption module, used for locally encrypting the face part in the face image data according to the local face coordinate data to obtain encrypted face image data; The decryption module is used to send the encrypted face image data to the target end for decryption.

6. The low-computing-power face encryption transmission system as claimed in claim 5, characterized in that: The face detection module includes a preprocessing unit and a detection unit; The preprocessing unit is used to perform normalization processing on the face image data to obtain preprocessed face image data; The detection unit is used to perform face detection of preset dimensions on the face part in the pre-processed face image data to obtain the local face coordinate data; wherein the local face coordinate data includes face frame center point data, face frame size data and face key point data.

7. The low-computing-power face encryption transmission system as claimed in claim 5, characterized in that: The local encryption module includes a serialization unit, a serial encryption unit and a replacement unit; The serialization unit is used to serialize the face part in the face image data according to the local face coordinate data to obtain plain text pixel data of the face image; The sequence encryption unit is used to perform sequence encryption on the plaintext pixel data of the face image based on an encryption algorithm to obtain ciphertext pixel data of the face image; The replacement unit is used to replace the face part in the face image data according to the face image ciphertext pixel data to obtain the encrypted face image data.

8. The low-computing-power face encryption transmission system as claimed in claim 7, characterized in that: The encryption algorithm is based on the AES algorithm and the SM4 algorithm.

9. A computer device, characterized in that: include: A memory, a processor, and a low-computing-power face encryption transmission program stored in the memory and executable on the processor, wherein the processor implements the steps in the low-computing-power face encryption transmission method as described in any one of claims 1 to 4 when executing the low-computing-power face encryption transmission program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a low-computing-power face encryption transmission program, and when the low-computing-power face encryption transmission program is executed by the processor, the steps in the low-computing-power face encryption transmission method as described in any one of claims 1-4 are implemented.

Images