Delivery signing method based on face recognition

Abstract

The invention discloses a delivery signing method based on face recognition, and relates to the technical field of image recognition. Comprising the following steps: performing face registration, and establishing a face data pre-processing a front face image uploaded by a retailer, generating a feature vector by adopting a deep learning algorithm according to the pre-processed face image, encrypting the feature vector, binding the encrypted feature vector with corresponding identity information, and storing the encrypted feature vector in the face database, a retailer face image is shot according to an order corresponding to a deliveryman, the shot retailer face image is preprocessed, a feature vector is extracted, an ArcFace face recognition algorithm is adopted to compare whether the similarity between the feature vector obtained by the shot retailer face image and the feature vector in a face database is larger than a threshold value or not, and if yes, delivery signing is carried out; if yes, the two feature vectors belong to the same face image, signature verification succeeds, otherwise, the two feature vectors do not belong to the same face image, and shooting verification is conducted again or a standby verification mode is adopted.

Description

Technical Field

[0001] The present invention discloses a delivery receipt method based on face recognition, which relates to the technical field of image recognition. Background Art

[0002] After a retailer places an order, the logistics center sorts the goods, and then the deliveryman delivers the packaged goods to the retailer. According to the delivery requirements, in addition to delivering the goods to the retailer's store or home, the deliveryman also needs to leave a trace to ensure that the goods are signed by the retailer himself / herself, so as to avoid disputes caused by signing on behalf of others.

[0003] Currently, for the supervision of the traceability and delivery handover specifications of goods, the goods are mainly signed by taking handover photos, having the retailer sign, scanning the retailer's QR code, etc. It is very difficult to identify whether the signer is the retailer himself / herself by the above methods, and the number of photos manually verified every day is huge, resulting in low efficiency. Summary of the Invention

[0004] In view of the problem of difficult screening of the identity of the signer during the delivery process, the present invention provides a delivery receipt method based on face recognition, which takes face recognition technology as the core and uses information technology means to identify the face identity in the photo in real time or with a delay, solves the problem of large workload in subsequent screening and reviewing photos, and helps to solve disputes caused by signing on behalf of others.

[0005] The specific solution proposed by the present invention is as follows:

[0006] The present invention provides a delivery receipt method based on face recognition, including:

[0007] Perform face registration and establish a face database: preprocess the frontal face image uploaded by the retailer, generate a feature vector according to the preprocessed face image by using a deep learning algorithm, encrypt the feature vector and bind it with the corresponding identity information, and save it to the face database.

[0008] Perform delivery receipt: take a face image of the retailer according to the order corresponding to the deliveryman, preprocess the taken face image of the retailer, extract the feature vector, and use the ArcFace face recognition algorithm to compare whether the similarity between the feature vector obtained from the taken face image of the retailer and the feature vector in the face database is greater than the threshold. If so, the two feature vectors belong to the same face image and the signature verification is successful; otherwise, they do not belong to the same face image, and re-take the photo for verification or use an alternative verification method.

[0009] Furthermore, the face registration in the delivery receipt method based on face recognition includes:

[0010] Preprocess the frontal face image uploaded by the retailer by performing light correction and image cropping.

[0011] Generate a feature vector from the pre - processed face image using a Convolutional Neural Network (CNN).

[0012] Encrypt the feature vector using the AES - 256 or the national cryptographic SM4 algorithm, bind it with the corresponding identity information, and save it to the face database.

[0013] Furthermore, for the delivery receipt signing of the described method for delivery receipt signing based on face recognition, it includes:

[0014] Perform pre - processing of light correction and image cropping on the captured face image of the retailer.

[0015] Use a Deep Convolutional Neural Network (DCNN) to extract the features of the pre - processed face image. Among them, use the ResNet network structure, add a fully - connected layer before the last layer of the network, and output the face feature vector.

[0016] Normalize each dimension of the feature vector.

[0017] Use the ArcFace face recognition algorithm to compare whether the similarity between the feature vector obtained from the captured face image of the retailer and the feature vector in the face database is greater than the threshold. The preset threshold is 0.9. If so, the two feature vectors belong to the same face image, and the signature verification is successful; otherwise, they do not belong to the same face image, and re - capture verification or an alternative verification method is performed.

[0018] Furthermore, after the signature verification of the described method for delivery receipt signing based on face recognition is successful, generate a receipt record, add a timestamp and a digital signature, and use the SHA - 256 hashing algorithm to prevent tampering with the receipt record.

[0019] The present invention also provides a delivery receipt signing device based on face recognition, including a face registration module, a face recognition module, and a receipt confirmation module.

[0020] The face registration module performs face registration and establishes a face database: perform pre - processing on the frontal face image uploaded by the retailer, generate a feature vector from the pre - processed face image using a deep learning algorithm, encrypt the feature vector and bind it with the corresponding identity information, and save it to the face database.

[0021] When performing delivery receipt signing, the face recognition module captures the face image of the retailer according to the corresponding order of the deliveryman, preprocesses the captured face image of the retailer, extracts the feature vector, and uses the ArcFace face recognition algorithm to compare whether the similarity between the feature vector obtained from the captured face image of the retailer and the feature vector in the face database is greater than the threshold. If so, the two feature vectors belong to the same face image, and the signature verification module successfully verifies the signature. Otherwise, they do not belong to the same face image, and the face recognition module performs re-capture verification.

[0022] Furthermore, the face registration module of the delivery receipt signing device based on face recognition performs face registration, including:

[0023] Perform preprocessing of light correction and image cropping on the frontal face image uploaded by the retailer.

[0024] Generate a feature vector using the convolutional neural network CNN based on the preprocessed face image.

[0025] Encrypt the feature vector using the AES-256 or national cipher SM4 algorithm, bind it with the corresponding identity information, and save it to the face database.

[0026] Furthermore, when the delivery receipt signing device based on face recognition performs delivery receipt signing, the face recognition module performs preprocessing of light correction and image cropping on the captured face image of the retailer.

[0027] The face recognition module uses the deep convolutional neural network DCNN to extract the features of the preprocessed face image. Among them, the ResNet network structure is used, and a fully connected layer is added before the last layer of the network to output the face feature vector.

[0028] The face recognition module normalizes each dimension of the feature vector.

[0029] The face recognition module uses the ArcFace face recognition algorithm to compare whether the similarity between the feature vector obtained from the captured face image of the retailer and the feature vector in the face database is greater than the threshold. The preset threshold is 0.9. If so, the two feature vectors belong to the same face image, and the signature verification is successful. Otherwise, they do not belong to the same face image, and re-capture verification or an alternative verification method is performed.

[0030] Furthermore, after the signature verification module of the delivery receipt signing device based on face recognition successfully verifies the signature, it generates a receipt record and adds a timestamp and a digital signature, and uses the SHA-256 hashing algorithm to prevent the receipt record from being tampered with.

[0031] The advantages of the present invention are:

[0032] During the delivery process, with face recognition technology at the core, the identity of the face in the photo is identified in real time or with a time delay by means of informatization, ensuring that the retailer signs for the goods himself, ensuring home delivery, and providing effective data support for internal industry supervision and evidence sampling. It solves the problem of the large workload of screening and reviewing photos later and helps to resolve disputes caused by non-personal signing. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a schematic diagram of the face registration implementation process of the present invention.

[0034] Figure 2 It is a schematic diagram of the face recognition implementation process. DETAILED DESCRIPTION OF THE INVENTION

[0035] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, so that those skilled in the art can better understand the present invention and be able to implement it, but the specific embodiments cited are not intended to limit the present invention.

[0036] Embodiment 1

[0037] The present invention provides a delivery receipt method based on face recognition, including:

[0038] Perform face registration and establish a face database: preprocess the frontal face image uploaded by the retailer, generate a feature vector using a deep learning algorithm based on the preprocessed face image, encrypt the feature vector and bind it to the corresponding identity information, and save it to the face database.

[0039] It may include: preprocessing the frontal face image uploaded by the retailer by performing light correction and image cropping, and generating a feature vector using a convolutional neural network CNN based on the preprocessed face image,

[0040] Encrypt the feature vector using the AES-256 or national cipher SM4 algorithm, bind it to the corresponding identity information, and save it to the face database.

[0041] Perform delivery receipt: take a face image of the retailer according to the order corresponding to the deliveryman, preprocess the taken face image of the retailer, extract the feature vector, and use the ArcFace face recognition algorithm to compare whether the similarity between the feature vector obtained from the taken face image of the retailer and the feature vector in the face database is greater than the threshold. If so, the two feature vectors belong to the same face image and the signature verification is successful; otherwise, they do not belong to the same face image, and re-take the verification or use an alternative verification method.

[0042] Wherein the performing of the delivery receipt includes: preprocessing the taken face image of the retailer by performing light correction and image cropping,

[0043] Use a deep convolutional neural network (DCNN) to extract the features of the preprocessed face image. Among them, the ResNet network structure is used, and a fully connected layer is added before the last layer of the network to output the face feature vector. Normalize each dimension of the feature vector.

[0044] Adopt the ArcFace face recognition algorithm to compare whether the similarity between the feature vector obtained from the photographed face image of the retailer and the feature vector in the face database is greater than the threshold. The preset threshold is 0.9. If so, the two feature vectors belong to the same face image, and the signature verification is successful. Otherwise, they do not belong to the same face image, and re-shooting verification or an alternative verification method is adopted.

[0045] After the signature verification is successful, generate a receipt record and add a timestamp and a digital signature, and use the SHA-256 hashing algorithm to prevent the receipt record from being tampered with.

[0046] Embodiment 2

[0047] The present invention also provides a delivery receipt device based on face recognition, including a face registration module, a face recognition module, and a receipt confirmation module.

[0048] The face registration module performs face registration and establishes a face database: preprocess the frontal face image uploaded by the retailer, generate a feature vector according to the preprocessed face image using a deep learning algorithm, encrypt the feature vector and bind it to the corresponding identity information, and save it to the face database.

[0049] When performing delivery receipt, the face recognition module takes a face image of the retailer according to the corresponding order of the deliveryman, preprocesses the taken face image of the retailer, extracts the feature vector, and adopts the ArcFace face recognition algorithm to compare whether the similarity between the feature vector obtained from the taken face image of the retailer and the feature vector in the face database is greater than the threshold. If so, the two feature vectors belong to the same face image, and the receipt confirmation module has a successful signature verification. Otherwise, they do not belong to the same face image, and the face recognition module performs re-shooting verification.

[0050] Regarding the information interaction and execution process among the above-mentioned modules, since they are based on the same concept as the method embodiment of the present invention, the specific content can be referred to the description in the method embodiment of the present invention, and will not be elaborated here.

[0051] Similarly, in the distribution process of the device of the present invention, with face recognition technology as the core, it can identify the face identity in the photo in real time or with a delay by means of informatization, ensuring that the retailer signs for the receipt himself, ensuring door-to-door delivery, and providing effective data support for internal industry supervision and evidence sampling. It solves the problem of large workload in subsequent screening and reviewing photos, and helps to solve disputes caused by non-personal receipt.

[0052] It should be noted that not all steps and modules in the above-mentioned processes and device structures are necessary, and some steps or modules can be ignored according to actual needs. The execution order of each step is not fixed and can be adjusted as required. The system structures described in the above embodiments can be physical structures or logical structures, that is, some modules may be implemented by the same physical entity, or some modules may be implemented separately by multiple physical entities, or they can be jointly implemented by some components in multiple independent devices.

[0053] The above-mentioned embodiments are only preferred embodiments cited to fully illustrate the present invention, and the protection scope of the present invention is not limited thereto. Equivalent substitutions or transformations made by those skilled in the art on the basis of the present invention are within the protection scope of the present invention. The protection scope of the present invention is subject to the claims.

Claims

1. A delivery signing method based on face recognition, characterized by include: Perform face registration and establish a face database: Pre-process the frontal face images uploaded by retailers, use a deep learning algorithm to generate feature vectors based on the pre-processed face images, encrypt the feature vectors and bind them to the corresponding identity information, and save them to the face database. Signing for delivery: The deliveryman takes a photo of the retailer's face image according to the corresponding order, pre-processes the photographed retailer's face image, extracts the feature vector, and uses the ArcFace face recognition algorithm to compare the similarity between the feature vector obtained from the photographed retailer's face image and the feature vector in the face database to see if it is greater than a threshold. If so, the two feature vectors belong to the same face image and the signature is successfully verified. Otherwise, they do not belong to the same face image and the verification is performed again or an alternative verification method is used.

2. According to the method for delivery and receipt based on face recognition according to claim 1, it is characterized by: The face registration includes: Perform light correction and image cropping preprocessing on the frontal face images uploaded by retailers. According to the preprocessed face image, a convolutional neural network CNN is used to generate feature vectors. The feature vector is encrypted using AES-256 or the national secret SM4 algorithm, bound to the corresponding identity information, and saved in the face database.

3. According to the method for delivery and receipt based on face recognition according to claim 1, it is characterized by: The delivery and receipt signing includes: Perform light correction and image cropping preprocessing on the facial images of retailers. A deep convolutional neural network (DCNN) is used to extract the features of the preprocessed face image. A ResNet network structure is used, and a fully connected layer is added before the last layer of the network to output the face feature vector. Normalize each dimension of the feature vector. The ArcFace facial recognition algorithm is used to compare the similarity between the feature vector obtained from the photographed face image of the retailer and the feature vector in the face database to see if it is greater than a threshold. The preset threshold is 0.

9. If so, the two feature vectors belong to the same face image and the signature verification is successful. Otherwise, they do not belong to the same face image and a re-photographing verification is performed or an alternative verification method is used.

4. A delivery receipt method based on face recognition according to claim 3, characterized in that After the signature verification is successful, a receipt record is generated and a timestamp and digital signature are added. The SHA-256 hash algorithm is used to prevent tampering with the receipt record.

5. A delivery and receipt device based on face recognition, characterized in that Including face registration module, face recognition module and receipt confirmation module, The face registration module performs face registration and establishes a face database: it pre-processes the frontal face images uploaded by retailers, generates feature vectors based on the pre-processed face images using a deep learning algorithm, encrypts the feature vectors and binds them to the corresponding identity information, and saves them to the face database. When signing for delivery, the facial recognition module takes a photo of the retailer's facial image based on the deliveryman's corresponding order, preprocesses the photographed facial image, extracts feature vectors, and uses the ArcFace facial recognition algorithm to compare whether the similarity between the feature vector obtained from the photographed facial image of the retailer and the feature vector in the face database is greater than a threshold. If so, the two feature vectors belong to the same facial image and the signing confirmation module verifies the signature successfully. Otherwise, they do not belong to the same facial image and the facial recognition module re-takes the photo for verification.

6. The delivery and receipt device based on face recognition according to claim 5 is characterized in that The face registration module performs face registration, including: Perform light correction and image cropping preprocessing on the frontal face images uploaded by retailers. According to the preprocessed face image, a convolutional neural network CNN is used to generate feature vectors. The feature vector is encrypted using AES-256 or the national secret SM4 algorithm, bound to the corresponding identity information, and saved in the face database.

7. The delivery and receipt device based on face recognition according to claim 5 is characterized in that When signing for delivery, the facial recognition module performs light correction and image cropping pre-processing on the retailer's facial image. The face recognition module uses a deep convolutional neural network (DCNN) to extract the features of the preprocessed face image. The ResNet network structure is used, and a fully connected layer is added before the last layer of the network to output the face feature vector. The face recognition module normalizes each dimension of the feature vector. The face recognition module uses the ArcFace face recognition algorithm to compare the similarity between the feature vector obtained from the photographed face image of the retailer and the feature vector in the face database to see if it is greater than a threshold. The preset threshold is 0.

9. If so, the two feature vectors belong to the same face image and the signature verification is successful. Otherwise, they do not belong to the same face image and a re-photographing verification is performed or an alternative verification method is used.

8. The delivery and receipt device based on face recognition according to claim 7 is characterized in that After the signature is successfully verified by the receipt confirmation module, a receipt record is generated and a timestamp and digital signature are added. The SHA-256 hash algorithm is used to prevent tampering with the receipt record.

Images