A charging pile non-inductive charging method and system

Abstract

The application provides a charging pile non-inductive charging method and system in the technical field of charging piles, and the method comprises the following steps: step S1, creating a face recognition model; step S2, obtaining a large number of historical face images to construct a data set, training the face recognition model through the data set, and deploying the trained face recognition model to the charging pile; step S3, positioning the mobile terminal on the electric vehicle by the charging pile to obtain the pile-vehicle distance, and performing primary verification based on the pile-vehicle distance; step S4, during the process of inserting the charging gun into the charging port of the electric vehicle, the charging pile collects real-time face images, inputs the real-time face images into the face recognition model to obtain face recognition results, and performs secondary verification based on the face recognition results; and step S5, after the charging gun is recognized to be inserted, the charging gun interacts with the electric vehicle to perform tertiary verification, and the charging authorization is started after the verification is passed. The application has the advantages that the safety and convenience of electric vehicle charging are greatly improved.

Description

Technical Field

[0001] This invention relates to the field of charging pile technology, and in particular to a method and system for contactless charging of charging piles. Background Technology

[0002] A charging station is a device that provides electrical energy to electric vehicles, enabling them to store enough electricity to support their operation. With the rapid development of electric vehicles, the demand for charging stations is increasing daily. Charging stations can be classified into AC charging stations (slow charging) and DC charging stations (fast charging) based on their charging method. AC charging stations directly supply AC power from the grid to the electric vehicle's onboard charger, which converts the AC power into DC power to charge the battery. DC charging stations, on the other hand, directly supply DC power to the electric vehicle's battery, resulting in faster charging. AC charging stations are generally used by individual users within residential communities, while DC charging stations are typically used in charging stations.

[0003] Currently, there are several ways to start charging an electric vehicle at a charging station: plug-and-charge (charging begins as soon as the charging gun is inserted into the charging port), card-swipe charging (the user needs to swipe a card to start charging after the charging gun is inserted into the charging port), and remote-controlled charging (the user needs to use an app to start charging after the charging gun is inserted into the charging port). In terms of convenience, plug-and-charge is the most convenient method, requiring no additional user intervention. However, plug-and-charge has lower security, as anyone can use the charging station. Card-swipe charging and remote-controlled charging, on the other hand, are cumbersome and detrimental to the user experience.

[0004] Therefore, how to provide a seamless charging method and system for charging piles to improve the safety and convenience of electric vehicle charging has become an urgent technical problem to be solved. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method and system for contactless charging of charging piles, so as to improve the safety and convenience of electric vehicle charging.

[0006] In a first aspect, the present invention provides a method for contactless charging of a charging pile, comprising the following steps:

[0007] Step S1: Create a face recognition model and set the loss function of the face recognition model;

[0008] Step S2: Obtain a large number of historical face images, preprocess each of the historical face images to construct a dataset, train the face recognition model using the dataset, and deploy the trained face recognition model to the charging pile.

[0009] Step S3: The charging pile locates the mobile terminal on the electric vehicle to obtain the distance between the charging pile and the vehicle, and performs a first-level verification based on the distance between the charging pile and the vehicle.

[0010] Step S4: During the process of inserting the charging gun into the charging port of the electric vehicle, the charging pile collects real-time facial images, inputs the real-time facial images into the facial recognition model to obtain facial recognition results, and performs secondary verification based on the facial recognition results.

[0011] Step S5: After the charging gun recognizes that the charging gun is plugged in, it interacts with the electric vehicle to perform a three-level verification. If the verification is successful, charging authorization is initiated, and then the electric vehicle is charged.

[0012] Step S6: During the charging process of the charging pile to the electric vehicle, the charging strategy is dynamically adjusted until charging is completed and the charging authorization is closed.

[0013] Step S7: The charging pile records the charging log and stores the charging log to the blockchain.

[0014] Furthermore, step S1 specifically includes:

[0015] A face recognition model is created based on an image preprocessing module, a face feature extraction module, a feature comparison module, and a recognition module. The loss function of the face recognition model is set as Softmax Loss.

[0016] The image preprocessing module, face feature extraction module, feature comparison module, and recognition module are connected sequentially. The image preprocessing module performs image normalization, size adjustment, and noise reduction on the input face image, and inputs the preprocessed face image into the face feature extraction module. The face feature extraction module extracts face features from the face image through convolutional layers, pooling layers, and activation functions, and inputs the face features into the feature comparison module. The feature comparison module calculates the cosine similarity between the face features and the face features registered in the database, and inputs the cosine similarity into the recognition module. The recognition module outputs the face recognition result based on the cosine similarity.

[0017] Furthermore, step S2 specifically includes:

[0018] A large number of historical face images are acquired, and each of the historical face images is preprocessed, including at least cropping, resizing, illumination correction, grayscale conversion, normalization and noise reduction. After labeling the names of the people in each of the preprocessed historical face images, a dataset is constructed.

[0019] The dataset is divided into a training set, a validation set, and a test set in a 7:2:1 ratio. The face recognition model is trained on the training set until the loss function's loss value is less than a preset loss threshold. During training, dynamic pruning and knowledge distillation techniques are used to compress the face recognition model. Next, the trained face recognition model is validated on the validation set to determine if the recognition accuracy is greater than a preset accuracy threshold. If not, the validation fails, and the dataset is expanded for continued training; if yes, the validation succeeds, and...

[0020] The successfully validated face recognition model is tested using the test set to determine whether the recognition confidence level is greater than a preset confidence threshold. If not, the test fails, and the dataset is expanded to continue training; if yes, the test succeeds, training ends, and the trained face recognition model is deployed to the charging station.

[0021] Step S3 specifically involves:

[0022] The charging pile uses dual-mode positioning technology of UWB and Bluetooth to locate the mobile terminal on the electric vehicle, obtain the positioning point, and compensate the positioning point by combining the RSS I of the mobile terminal. Based on the positioning point, the distance between the charging pile and the vehicle is calculated, and the distance between the charging pile and the vehicle is verified at the first level based on a preset distance threshold.

[0023] Furthermore, step S4 specifically includes:

[0024] During the process of inserting the charging gun into the charging port of the electric vehicle, the charging pile captures real-time facial images through a camera. After preprocessing the real-time facial images, they are input into a facial recognition model. The facial recognition model performs matching and recognition based on a pre-stored user management table to obtain facial recognition results. Secondary verification is then performed based on the facial recognition results.

[0025] Step S5 specifically involves:

[0026] After the charging gun recognizes that the charging gun is plugged in, it establishes a handshake communication with the electric vehicle, then interacts with the electric vehicle to obtain its operating status. Based on the operating status, it performs a three-level verification. If the verification passes, it initiates charging authorization and then charges the electric vehicle.

[0027] Furthermore, step S6 specifically includes:

[0028] During the charging process of electric vehicles, the charging pile dynamically adjusts the charging strategy based on historical charging data, real-time electricity price and grid load until the preset charging cutoff conditions are met to complete the charging and close the charging authorization.

[0029] Step S7 specifically involves:

[0030] The charging station records charging logs in real time, including at least charging time, charging amount, electricity price, charging power, account, vehicle model, and vehicle serial number. The charging logs are encrypted using the SM4 algorithm to obtain a first layer of encrypted text. A 16-bit random string is generated and inserted into the beginning, middle, and end of the first layer of encrypted text to obtain a second layer of encrypted text. The second layer of encrypted text is then encrypted into a third layer of encrypted text using the IDEA algorithm. The numbers 0 and 0, 1 and 1, and 5 and 5 in the third layer of encrypted text are swapped to obtain a fourth layer of encrypted text. The fourth layer of encrypted text is then encrypted using the ECDSA algorithm to obtain an encrypted log, which is then stored on the blockchain.

[0031] Secondly, the present invention provides a contactless charging system for charging piles, comprising the following modules:

[0032] The face recognition model creation module is used to create a face recognition model and set the loss function of the face recognition model;

[0033] The face recognition model deployment module is used to acquire a large number of historical face images, preprocess each of the historical face images to construct a dataset, train the face recognition model using the dataset, and deploy the trained face recognition model to the charging pile.

[0034] The charging pile-vehicle distance calculation module is used to locate the mobile terminal on the electric vehicle by the charging pile, thereby obtaining the charging pile-vehicle distance, and performing a first-level verification based on the charging pile-vehicle distance;

[0035] The face recognition module is used to collect real-time face images during the process of inserting the charging gun into the charging port of an electric vehicle, input the real-time face images into the face recognition model to obtain face recognition results, and perform secondary verification based on the face recognition results;

[0036] The vehicle status verification module is used to interact with the electric vehicle to perform a three-level verification after the charging gun recognizes that the charging gun is plugged in. If the verification is successful, charging authorization is initiated, and then the electric vehicle is charged.

[0037] The charging module is used to dynamically adjust the charging strategy during the charging process of electric vehicles by the charging pile until charging is completed and the charging authorization is closed.

[0038] The charging log management module is used to record charging logs at the charging pile and store the charging logs on the blockchain.

[0039] Furthermore, the face recognition model creation module is specifically used for:

[0040] A face recognition model is created based on an image preprocessing module, a face feature extraction module, a feature comparison module, and a recognition module. The loss function of the face recognition model is set as Softmax Loss.

[0041] The image preprocessing module, face feature extraction module, feature comparison module, and recognition module are connected sequentially. The image preprocessing module performs image normalization, size adjustment, and noise reduction on the input face image, and inputs the preprocessed face image into the face feature extraction module. The face feature extraction module extracts face features from the face image through convolutional layers, pooling layers, and activation functions, and inputs the face features into the feature comparison module. The feature comparison module calculates the cosine similarity between the face features and the face features registered in the database, and inputs the cosine similarity into the recognition module. The recognition module outputs the face recognition result based on the cosine similarity.

[0042] Furthermore, the face recognition model deployment module is specifically used for:

[0043] A large number of historical face images are acquired, and each of the historical face images is preprocessed, including at least cropping, resizing, illumination correction, grayscale conversion, normalization and noise reduction. After labeling the names of the people in each of the preprocessed historical face images, a dataset is constructed.

[0044] The dataset is divided into a training set, a validation set, and a test set in a 7:2:1 ratio. The face recognition model is trained on the training set until the loss function's loss value is less than a preset loss threshold. During training, dynamic pruning and knowledge distillation techniques are used to compress the face recognition model. Next, the trained face recognition model is validated on the validation set to determine if the recognition accuracy is greater than a preset accuracy threshold. If not, the validation fails, and the dataset is expanded for continued training; if yes, the validation succeeds, and...

[0045] The successfully validated face recognition model is tested using the test set to determine whether the recognition confidence level is greater than a preset confidence threshold. If not, the test fails, and the dataset is expanded to continue training; if yes, the test succeeds, training ends, and the trained face recognition model is deployed to the charging station.

[0046] The pile driver distance calculation module is specifically used for:

[0047] The charging pile uses dual-mode positioning technology of UWB and Bluetooth to locate the mobile terminal on the electric vehicle, obtain the positioning point, and compensate the positioning point by combining the RSS I of the mobile terminal. Based on the positioning point, the distance between the charging pile and the vehicle is calculated, and the distance between the charging pile and the vehicle is verified at the first level based on a preset distance threshold.

[0048] Furthermore, the face recognition module is specifically used for:

[0049] During the process of inserting the charging gun into the charging port of the electric vehicle, the charging pile captures real-time facial images through a camera. After preprocessing the real-time facial images, they are input into a facial recognition model. The facial recognition model performs matching and recognition based on a pre-stored user management table to obtain facial recognition results. Secondary verification is then performed based on the facial recognition results.

[0050] The vehicle status verification module is specifically used for:

[0051] After the charging gun recognizes that the charging gun is plugged in, it establishes a handshake communication with the electric vehicle, then interacts with the electric vehicle to obtain its operating status. Based on the operating status, it performs a three-level verification. If the verification passes, it initiates charging authorization and then charges the electric vehicle.

[0052] Furthermore, the charging module is specifically used for:

[0053] During the charging process of electric vehicles, the charging pile dynamically adjusts the charging strategy based on historical charging data, real-time electricity price and grid load until the preset charging cutoff conditions are met to complete the charging and close the charging authorization.

[0054] The charging log management module is specifically used for:

[0055] The charging station records charging logs in real time, including at least charging time, charging amount, electricity price, charging power, account, vehicle model, and vehicle serial number. The charging logs are encrypted using the SM4 algorithm to obtain a first layer of encrypted text. A 16-bit random string is generated and inserted into the beginning, middle, and end of the first layer of encrypted text to obtain a second layer of encrypted text. The second layer of encrypted text is then encrypted into a third layer of encrypted text using the IDEA algorithm. The numbers 0 and 0, 1 and 1, and 5 and 5 in the third layer of encrypted text are swapped to obtain a fourth layer of encrypted text. The fourth layer of encrypted text is then encrypted using the ECDSA algorithm to obtain an encrypted log, which is then stored on the blockchain.

[0056] The advantages of this invention are:

[0057] 1. By creating a face recognition model and setting a loss function, a large number of historical face images are acquired. After preprocessing each historical face image, a dataset is constructed. The face recognition model is trained using this dataset and then deployed to charging stations. The charging station locates the mobile terminal on the electric vehicle to obtain the distance between the charging station and the vehicle, and performs a first-level verification based on this distance. While the charging gun is being inserted into the charging port of the electric vehicle, the charging station collects real-time face images and inputs these images into the face recognition model to obtain the face recognition result, which is then used for a second-level verification. After the charging gun is detected as inserted, it interacts with the electric vehicle for a third-level verification. If the verification is successful, charging authorization is initiated, and the electric vehicle is then charged. During the charging process, the charging station dynamically adjusts the charging strategy until charging is complete, and then closes the charging authorization. The charging station records the charging log and stores it on the blockchain. In other words, the charging station performs a triple verification based on the distance between the charging station and the vehicle, the face recognition result, and the operating status of the electric vehicle. Charging is only initiated after the verification is successful. The entire process is completed automatically during the user's parking and charging process, without any additional user intervention, greatly improving the safety and convenience of electric vehicle charging.

[0058] 2. By performing preprocessing on each historical face image, including at least cropping, resizing, illumination correction, grayscale conversion, normalization, and noise reduction, the quality of the dataset is effectively improved, thereby greatly enhancing the training effect of the face recognition model.

[0059] 3. By dividing the dataset into training, validation, and test sets, the face recognition model is trained on the training set until the loss function value is less than a preset loss threshold. During training, dynamic pruning and knowledge distillation techniques are used to compress the face recognition model. Then, the trained face recognition model is validated on the validation set, and the validated face recognition model is tested on the test set. After successful testing, the face recognition model is deployed to charging stations. In other words, the face recognition model is continuously validated and tested during training, which greatly improves the face recognition accuracy. Furthermore, the dynamic pruning and knowledge distillation techniques effectively reduce the size of the face recognition model, making it easier to deploy on resource-constrained charging stations.

[0060] 4. The mobile terminal on the electric vehicle is located using dual-mode positioning technology of UWB and Bluetooth to obtain the positioning point. After compensating the positioning point with the RSSI of the mobile terminal, the distance between the charging pile and the vehicle is calculated based on the positioning point. That is, the calculation of the distance between the charging pile and the vehicle combines UWB, Bluetooth and RSSI, which greatly improves the accuracy of the distance calculation.

[0061] 5. By dynamically adjusting the charging strategy based on historical charging data, real-time electricity prices, and grid load during the charging process of electric vehicles at charging stations, the charging costs and the burden on charging stations can be effectively balanced.

[0062] 6. By recording charging logs in real time, including at least charging time, charging amount, electricity price, charging power, account, vehicle model, and vehicle serial number, it is easy to trace the source later.

[0063] 7. The charging log is encrypted using the SM4 algorithm to obtain a first layer of encrypted text. A 16-bit random string is generated and inserted into the beginning, middle, and end of the first layer of encrypted text to obtain a second layer of encrypted text. The second layer of encrypted text is then encrypted into a third layer of encrypted text using the IDEA algorithm. The numbers 0 and 0, 1 and 1, and 5 and 5 in the third layer of encrypted text are swapped to obtain a fourth layer of encrypted text. The fourth layer of encrypted text is then encrypted using the ECDSA algorithm to obtain an encrypted log, which is then stored on the blockchain. In other words, the encryption process of the charging log combines triple encryption (SM4 symmetric encryption, IDEA symmetric encryption, and ECDSA asymmetric encryption) with second-order data transformation (random string insertion and character replacement). Inserting random strings enhances resistance to replay attacks. The combination of national cryptographic algorithms and international standard algorithms forms a hybrid encryption system. The final encrypted log requires mastery of both the encryption algorithm and the data transformation rules to decrypt. Furthermore, the blockchain effectively prevents tampering, avoiding the theft and alteration of the charging log in plaintext, thus greatly improving the security of the charging log storage. Attached Figure Description

[0064] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0065] Figure 1 This is a flowchart of a contactless charging method for a charging pile according to the present invention.

[0066] Figure 2 This is a schematic diagram of the structure of a contactless charging system for a charging pile according to the present invention. Detailed Implementation

[0067] The technical solution in this application embodiment has the following general idea: triple verification is performed based on the distance between the charging station and the vehicle, the facial recognition result, and the operating status of the electric vehicle. The electric vehicle is only charged after the verification is passed. The whole process does not require any additional operation from the user and is completed automatically when the user parks and plugs in the charging gun, thereby improving the safety and convenience of electric vehicle charging.

[0068] Please refer to Figures 1 to 2 As shown, a preferred embodiment of the contactless charging method for charging piles of the present invention includes the following steps:

[0069] Step S1: Create a face recognition model and set the loss function of the face recognition model;

[0070] Step S2: Obtain a large number of historical face images, preprocess each of the historical face images to construct a dataset, train the face recognition model using the dataset, and deploy the trained face recognition model to the charging pile.

[0071] Step S3: The charging pile locates the mobile terminal on the electric vehicle to obtain the distance between the charging pile and the vehicle, and performs a first-level verification based on the distance between the charging pile and the vehicle.

[0072] Step S4: During the process of inserting the charging gun into the charging port of the electric vehicle, the charging pile acquires real-time facial images, inputs the real-time facial images into the facial recognition model to obtain facial recognition results, and performs secondary verification based on the facial recognition results; combining the facial recognition model with real-time image acquisition, contactless identity verification is achieved, avoiding the risk of loss of traditional physical media (such as RFID cards);

[0073] Step S5: After the charging gun recognizes that the charging gun is plugged in, it interacts with the electric vehicle to perform a three-level verification. If the verification is successful, charging authorization is initiated, and the electric vehicle is charged. A composite safety protection system is built through the three-level verification mechanism (charging station distance verification + real-time face recognition + operation status verification), which effectively improves the safety of electric vehicle charging.

[0074] Step S6: During the charging process of the charging pile to the electric vehicle, the charging strategy is dynamically adjusted until charging is completed and the charging authorization is closed.

[0075] Step S7: The charging station records charging logs and stores these logs on the blockchain. Blockchain technology solidifies the charging logs, ensuring their immutability and providing a reliable source for billing disputes.

[0076] This invention automates the entire process from the electric vehicle approaching to the completion of charging, requiring only the insertion of a charging gun, aligning with the future trend of smart transportation.

[0077] Step S1 specifically involves:

[0078] A face recognition model is created based on an image preprocessing module, a face feature extraction module, a feature comparison module, and a recognition module. The loss function of the face recognition model is set as Softmax Loss.

[0079] The image preprocessing module, face feature extraction module, feature comparison module, and recognition module are connected sequentially. The image preprocessing module performs image normalization, size adjustment, and noise reduction on the input face image, and inputs the preprocessed face image into the face feature extraction module. The face feature extraction module extracts face features from the face image through convolutional layers, pooling layers, and activation functions, and inputs the face features into the feature comparison module. The feature comparison module calculates the cosine similarity between the face features and the face features registered in the database, and inputs the cosine similarity into the recognition module. The recognition module outputs the face recognition result based on the cosine similarity.

[0080] Step S2 specifically involves:

[0081] A large number of historical face images are acquired, and each of the historical face images is preprocessed, including at least cropping, resizing, illumination correction, grayscale conversion, normalization and noise reduction. After labeling the names of the people in each of the preprocessed historical face images, a dataset is constructed.

[0082] By performing preprocessing on each historical face image, including at least cropping, resizing, illumination correction, grayscale conversion, normalization, and noise reduction, the quality of the dataset is effectively improved, thereby greatly enhancing the training effect of the face recognition model.

[0083] The dataset is divided into training, validation, and test sets in a 7:2:1 ratio. The face recognition model is trained on the training set until the loss function's loss value is less than a preset loss threshold. During training, dynamic pruning and knowledge distillation techniques are used to compress the face recognition model. Based on dynamic pruning and knowledge distillation, the lightweight face recognition model achieves a 40% reduction in size while maintaining 98.7% recognition accuracy. The trained face recognition model is then validated on the validation set to determine if the recognition accuracy exceeds a preset accuracy threshold. If not, the validation fails, and the dataset is expanded for continued training; if yes, the validation succeeds.

[0084] The successfully validated face recognition model is tested using the test set to determine whether the recognition confidence level is greater than a preset confidence threshold. If not, the test fails, and the dataset is expanded to continue training; if yes, the test succeeds, training ends, and the trained face recognition model is deployed to the charging station.

[0085] By dividing the dataset into training, validation, and test sets, the face recognition model is trained on the training set until the loss function value is less than a preset loss threshold. During training, dynamic pruning and knowledge distillation techniques are used to compress the face recognition model. The trained face recognition model is then validated on the validation set, and the validated model is tested on the test set. Once the test is successful, the face recognition model is deployed to charging stations. In short, the face recognition model undergoes continuous validation and testing during training, significantly improving face recognition accuracy. Furthermore, the dynamic pruning and knowledge distillation techniques effectively reduce the model's size, facilitating deployment on resource-constrained charging stations.

[0086] Step S3 specifically involves:

[0087] The charging pile uses dual-mode positioning technology of UWB and Bluetooth to locate the mobile terminal on the electric vehicle and obtain the positioning point. After compensating the positioning point with the RSS I of the mobile terminal, the positioning deviation is corrected in real time based on the historical signal strength database. The distance between the charging pile and the vehicle is calculated based on the positioning point and a first-level verification is performed on the distance between the charging pile and the vehicle based on a preset distance threshold.

[0088] The mobile terminal on the electric vehicle is located by using dual-mode positioning technology of UWB and Bluetooth to obtain the positioning point. After compensating the positioning point with the RSS I of the mobile terminal, the distance between the charging pile and the vehicle is calculated based on the positioning point. That is, the calculation of the distance between the charging pile and the vehicle combines UWB, Bluetooth and RSS I, which greatly improves the accuracy of the distance calculation.

[0089] By innovatively employing UWB and Bluetooth dual-mode positioning technology combined with RSS I compensation algorithm, the positioning accuracy is effectively improved to the centimeter level.

[0090] Step S4 specifically involves:

[0091] During the process of inserting the charging gun into the charging port of the electric vehicle, the charging pile captures real-time facial images through a camera. After preprocessing the real-time facial images, they are input into a facial recognition model. The facial recognition model performs matching and recognition based on a pre-stored user management table to obtain facial recognition results. Secondary verification is then performed based on the facial recognition results.

[0092] Real-time facial recognition during the charging gun connection process ensures the operator's legitimacy and prevents the risk of unauthorized charging.

[0093] Step S5 specifically involves:

[0094] After the charging gun detects that it is plugged in, it initiates a handshake communication with the electric vehicle, then interacts with the vehicle to obtain its operating status. Based on this status, it performs a three-level verification. If the verification passes, it initiates charging authorization and begins charging the electric vehicle. This ensures that the charging activity occurs in the correct vehicle and that the equipment is functioning properly.

[0095] A progressive safety protection system is formed through a three-level verification mechanism (pile-vehicle distance verification, facial biometric recognition, and vehicle status interaction) to effectively prevent unauthorized charging.

[0096] Step S6 specifically involves:

[0097] During the charging process of electric vehicles, the charging pile dynamically adjusts the charging strategy based on historical charging data, real-time electricity price and grid load until the preset charging cutoff conditions are met to complete the charging and close the charging authorization. Through a three-dimensional decision model of historical charging data + real-time electricity price + grid load, it supports cost optimization strategy under time-of-use pricing and realizes intelligent scheduling of grid load peak shaving and valley filling.

[0098] By dynamically adjusting the charging strategy based on historical charging data, real-time electricity prices, and grid load during the charging process of electric vehicles at charging stations, the charging costs and the burden on charging stations can be effectively balanced.

[0099] Step S7 specifically involves:

[0100] The charging station records charging logs in real time, including at least charging time, charging amount, electricity price, charging power, account, vehicle model, and vehicle serial number. The charging logs are encrypted using the SM4 algorithm to obtain a first layer of encrypted text. A 16-bit random string is generated and inserted into the beginning, middle, and end of the first layer of encrypted text to obtain a second layer of encrypted text. The second layer of encrypted text is then encrypted into a third layer of encrypted text using the IDEA algorithm. The numbers 0 and 0, 1 and 1, and 5 and 5 in the third layer of encrypted text are swapped to obtain a fourth layer of encrypted text. The fourth layer of encrypted text is then encrypted using the ECDSA algorithm to obtain an encrypted log, which is then stored on the blockchain.

[0101] By recording charging logs in real time, including at least charging time, charging amount, electricity price, charging power, account, car model, and car serial number, it is easier to trace the source later.

[0102] The charging log is encrypted using the SM4 algorithm to obtain a first layer of encrypted text. A 16-bit random string is then generated and inserted into the beginning, middle, and end of the first layer of encrypted text to obtain a second layer of encrypted text. The second layer of encrypted text is then encrypted into a third layer of encrypted text using the IDEA algorithm. The numbers 0 and 'O', 1 and 'L', and 5 and 'S' in the third layer of encrypted text are swapped to obtain a fourth layer of encrypted text. This fourth layer of encrypted text is then encrypted using the ECDSA algorithm to obtain an encrypted log, which is then stored on the blockchain. In other words, the encryption process of the charging log combines triple encryption (SM4 symmetric encryption, IDEA symmetric encryption, and ECDSA asymmetric encryption) with second-order data transformation (random string insertion and character replacement). Inserting random strings enhances resistance to replay attacks. The combination of Chinese national cryptographic algorithms and international standard algorithms forms a hybrid encryption system. The final encrypted log requires mastery of both the encryption algorithm and the data transformation rules to decrypt. Furthermore, the blockchain effectively prevents tampering, avoiding the theft and alteration of the charging log in plaintext, thus greatly improving the security of the charging log storage.

[0103] A preferred embodiment of the present invention provides a contactless charging system for charging piles, comprising the following modules:

[0104] The face recognition model creation module is used to create a face recognition model and set the loss function of the face recognition model;

[0105] The face recognition model deployment module is used to acquire a large number of historical face images, preprocess each of the historical face images to construct a dataset, train the face recognition model using the dataset, and deploy the trained face recognition model to the charging pile.

[0106] The charging pile-vehicle distance calculation module is used to locate the mobile terminal on the electric vehicle by the charging pile, thereby obtaining the charging pile-vehicle distance, and performing a first-level verification based on the charging pile-vehicle distance;

[0107] The face recognition module is used to collect real-time face images during the process of inserting the charging gun into the charging port of an electric vehicle. The real-time face images are then input into a face recognition model to obtain face recognition results, and secondary verification is performed based on the face recognition results. By combining the face recognition model with real-time image acquisition, contactless identity verification is achieved, avoiding the risk of loss of traditional physical media (such as RFID cards).

[0108] The vehicle status verification module is used to interact with the electric vehicle to perform three-level verification after the charging gun recognizes that the charging gun is plugged in. If the verification is successful, charging authorization is initiated, and then the electric vehicle is charged. The three-level verification mechanism (charging station distance verification + real-time face recognition + operation status verification) builds a composite safety protection system, which effectively improves the safety of electric vehicle charging.

[0109] The charging module is used to dynamically adjust the charging strategy during the charging process of electric vehicles by the charging pile until charging is completed and the charging authorization is closed.

[0110] The charging log management module is used to record charging logs at charging stations and store these logs on the blockchain. Blockchain technology is used to solidify the charging logs, ensuring their immutability and providing reliable traceability for billing disputes.

[0111] This invention automates the entire process from the electric vehicle approaching to the completion of charging, requiring only the insertion of a charging gun, aligning with the future trend of smart transportation.

[0112] The face recognition model creation module is specifically used for:

[0113] A face recognition model is created based on an image preprocessing module, a face feature extraction module, a feature comparison module, and a recognition module. The loss function of the face recognition model is set as Softmax Loss.

[0114] The image preprocessing module, face feature extraction module, feature comparison module, and recognition module are connected sequentially. The image preprocessing module performs image normalization, size adjustment, and noise reduction on the input face image, and inputs the preprocessed face image into the face feature extraction module. The face feature extraction module extracts face features from the face image through convolutional layers, pooling layers, and activation functions, and inputs the face features into the feature comparison module. The feature comparison module calculates the cosine similarity between the face features and the face features registered in the database, and inputs the cosine similarity into the recognition module. The recognition module outputs the face recognition result based on the cosine similarity.

[0115] The face recognition model deployment module is specifically used for:

[0116] A large number of historical face images are acquired, and each of the historical face images is preprocessed, including at least cropping, resizing, illumination correction, grayscale conversion, normalization and noise reduction. After labeling each of the preprocessed historical face images with the names of the people, a dataset is constructed.

[0117] By performing preprocessing on each historical face image, including at least cropping, resizing, illumination correction, grayscale conversion, normalization, and noise reduction, the quality of the dataset is effectively improved, thereby greatly enhancing the training effect of the face recognition model.

[0118] The dataset is divided into training, validation, and test sets in a 7:2:1 ratio. The face recognition model is trained on the training set until the loss function's loss value is less than a preset loss threshold. During training, dynamic pruning and knowledge distillation techniques are used to compress the face recognition model. Based on dynamic pruning and knowledge distillation, the lightweight face recognition model achieves a 40% reduction in size while maintaining 98.7% recognition accuracy. The trained face recognition model is then validated on the validation set to determine if the recognition accuracy exceeds a preset accuracy threshold. If not, the validation fails, and the dataset is expanded for continued training; if yes, the validation succeeds.

[0119] The successfully validated face recognition model is tested using the test set to determine whether the recognition confidence level is greater than a preset confidence threshold. If not, the test fails, and the dataset is expanded to continue training; if yes, the test succeeds, training ends, and the trained face recognition model is deployed to the charging station.

[0120] By dividing the dataset into training, validation, and test sets, the face recognition model is trained on the training set until the loss function value is less than a preset loss threshold. During training, dynamic pruning and knowledge distillation techniques are used to compress the face recognition model. The trained face recognition model is then validated on the validation set, and the validated model is tested on the test set. Once the test is successful, the face recognition model is deployed to charging stations. In short, the face recognition model undergoes continuous validation and testing during training, significantly improving face recognition accuracy. Furthermore, the dynamic pruning and knowledge distillation techniques effectively reduce the model's size, facilitating deployment on resource-constrained charging stations.

[0121] The pile driver distance calculation module is specifically used for:

[0122] The charging pile uses dual-mode positioning technology of UWB and Bluetooth to locate the mobile terminal on the electric vehicle and obtain the positioning point. After compensating the positioning point with the RSS I of the mobile terminal, the positioning deviation is corrected in real time based on the historical signal strength database. The distance between the charging pile and the vehicle is calculated based on the positioning point and a first-level verification is performed on the distance between the charging pile and the vehicle based on a preset distance threshold.

[0123] The mobile terminal on the electric vehicle is located by using dual-mode positioning technology of UWB and Bluetooth to obtain the positioning point. After compensating the positioning point with the RSS I of the mobile terminal, the distance between the charging pile and the vehicle is calculated based on the positioning point. That is, the calculation of the distance between the charging pile and the vehicle combines UWB, Bluetooth and RSS I, which greatly improves the accuracy of the distance calculation.

[0124] By innovatively employing UWB and Bluetooth dual-mode positioning technology combined with RSS I compensation algorithm, the positioning accuracy is effectively improved to the centimeter level.

[0125] The face recognition module is specifically used for:

[0126] During the process of inserting the charging gun into the charging port of the electric vehicle, the charging pile captures real-time facial images through a camera. After preprocessing the real-time facial images, they are input into a facial recognition model. The facial recognition model performs matching and recognition based on a pre-stored user management table to obtain facial recognition results. Secondary verification is then performed based on the facial recognition results.

[0127] Real-time facial recognition during the charging gun connection process ensures the operator's legitimacy and prevents the risk of unauthorized charging.

[0128] The vehicle status verification module is specifically used for:

[0129] After the charging gun detects that it is plugged in, it initiates a handshake communication with the electric vehicle, then interacts with the vehicle to obtain its operating status. Based on this status, it performs a three-level verification. If the verification passes, it initiates charging authorization and begins charging the electric vehicle. This ensures that the charging activity occurs in the correct vehicle and that the equipment is functioning properly.

[0130] A progressive safety protection system is formed through a three-level verification mechanism (pile-vehicle distance verification, facial biometric recognition, and vehicle status interaction) to effectively prevent unauthorized charging.

[0131] The charging module is specifically used for:

[0132] During the charging process of electric vehicles, the charging pile dynamically adjusts the charging strategy based on historical charging data, real-time electricity price and grid load until the preset charging cutoff conditions are met to complete the charging and close the charging authorization. Through a three-dimensional decision model of historical charging data + real-time electricity price + grid load, it supports cost optimization strategy under time-of-use pricing and realizes intelligent scheduling of grid load peak shaving and valley filling.

[0133] By dynamically adjusting the charging strategy based on historical charging data, real-time electricity prices, and grid load during the charging process of electric vehicles at charging stations, the charging costs and the burden on charging stations can be effectively balanced.

[0134] The charging log management module is specifically used for:

[0135] The charging station records charging logs in real time, including at least charging time, charging amount, electricity price, charging power, account, vehicle model, and vehicle serial number. The charging logs are encrypted using the SM4 algorithm to obtain a first layer of encrypted text. A 16-bit random string is generated and inserted into the beginning, middle, and end of the first layer of encrypted text to obtain a second layer of encrypted text. The second layer of encrypted text is then encrypted into a third layer of encrypted text using the IDEA algorithm. The numbers 0 and 0, 1 and 1, and 5 and 5 in the third layer of encrypted text are swapped to obtain a fourth layer of encrypted text. The fourth layer of encrypted text is then encrypted using the ECDSA algorithm to obtain an encrypted log, which is then stored on the blockchain.

[0136] By recording charging logs in real time, including at least charging time, charging amount, electricity price, charging power, account, car model, and car serial number, it is easier to trace the source later.

[0137] The charging log is encrypted using the SM4 algorithm to obtain a first layer of encrypted text. A 16-bit random string is then generated and inserted into the beginning, middle, and end of the first layer of encrypted text to obtain a second layer of encrypted text. The second layer of encrypted text is then encrypted into a third layer of encrypted text using the IDEA algorithm. The numbers 0 and 'O', 1 and 'L', and 5 and 'S' in the third layer of encrypted text are swapped to obtain a fourth layer of encrypted text. This fourth layer of encrypted text is then encrypted using the ECDSA algorithm to obtain an encrypted log, which is then stored on the blockchain. In other words, the encryption process of the charging log combines triple encryption (SM4 symmetric encryption, IDEA symmetric encryption, and ECDSA asymmetric encryption) with second-order data transformation (random string insertion and character replacement). Inserting random strings enhances resistance to replay attacks. The combination of Chinese national cryptographic algorithms and international standard algorithms forms a hybrid encryption system. The final encrypted log requires mastery of both the encryption algorithm and the data transformation rules to decrypt. Furthermore, the blockchain effectively prevents tampering, avoiding the theft and alteration of the charging log in plaintext, thus greatly improving the security of the charging log storage.

[0138] In summary, the advantages of this invention are as follows:

[0139] 1. By creating a face recognition model and setting a loss function, a large number of historical face images are acquired. After preprocessing each historical face image, a dataset is constructed. The face recognition model is trained using this dataset and then deployed to charging stations. The charging station locates the mobile terminal on the electric vehicle to obtain the distance between the charging station and the vehicle, and performs a first-level verification based on this distance. While the charging gun is being inserted into the charging port of the electric vehicle, the charging station collects real-time face images and inputs these images into the face recognition model to obtain the face recognition result, which is then used for a second-level verification. After the charging gun is detected as inserted, it interacts with the electric vehicle for a third-level verification. If the verification is successful, charging authorization is initiated, and the electric vehicle is then charged. During the charging process, the charging station dynamically adjusts the charging strategy until charging is complete, and then closes the charging authorization. The charging station records the charging log and stores it on the blockchain. In other words, the charging station performs a triple verification based on the distance between the charging station and the vehicle, the face recognition result, and the operating status of the electric vehicle. Charging is only initiated after the verification is successful. The entire process is completed automatically during the user's parking and charging process, without any additional user intervention, greatly improving the safety and convenience of electric vehicle charging.

[0140] 2. By performing preprocessing on each historical face image, including at least cropping, resizing, illumination correction, grayscale conversion, normalization, and noise reduction, the quality of the dataset is effectively improved, thereby greatly enhancing the training effect of the face recognition model.

[0141] 3. By dividing the dataset into training, validation, and test sets, the face recognition model is trained on the training set until the loss function value is less than a preset loss threshold. During training, dynamic pruning and knowledge distillation techniques are used to compress the face recognition model. Then, the trained face recognition model is validated on the validation set, and the validated face recognition model is tested on the test set. After successful testing, the face recognition model is deployed to charging stations. In other words, the face recognition model is continuously validated and tested during training, which greatly improves the face recognition accuracy. Furthermore, the dynamic pruning and knowledge distillation techniques effectively reduce the size of the face recognition model, making it easier to deploy on resource-constrained charging stations.

[0142] 4. The mobile terminal on the electric vehicle is located using dual-mode positioning technology of UWB and Bluetooth to obtain the positioning point. After compensating the positioning point with the RSSI of the mobile terminal, the distance between the charging pile and the vehicle is calculated based on the positioning point. That is, the calculation of the distance between the charging pile and the vehicle combines UWB, Bluetooth and RSSI, which greatly improves the accuracy of the distance calculation.

[0143] 5. By dynamically adjusting the charging strategy based on historical charging data, real-time electricity prices, and grid load during the charging process of electric vehicles at charging stations, the charging costs and the burden on charging stations can be effectively balanced.

[0144] 6. By recording charging logs in real time, including at least charging time, charging amount, electricity price, charging power, account, vehicle model, and vehicle serial number, it is easy to trace the source later.

[0145] 7. The charging log is encrypted using the SM4 algorithm to obtain a first layer of encrypted text. A 16-bit random string is generated and inserted into the beginning, middle, and end of the first layer of encrypted text to obtain a second layer of encrypted text. The second layer of encrypted text is then encrypted into a third layer of encrypted text using the IDEA algorithm. The numbers 0 and 0, 1 and 1, and 5 and 5 in the third layer of encrypted text are swapped to obtain a fourth layer of encrypted text. The fourth layer of encrypted text is then encrypted using the ECDSA algorithm to obtain the encrypted log, which is then stored on the blockchain. In other words, the encryption process of the charging log combines triple encryption (SM4 symmetric encryption, IDEA symmetric encryption, and ECDSA asymmetric encryption) with second-order data transformation (random string insertion and character replacement). Inserting random strings enhances resistance to replay attacks. The combination of national cryptographic algorithms and international standard algorithms forms a hybrid encryption system. The final encrypted log requires mastery of both the encryption algorithm and the data transformation rules to decrypt. Furthermore, the blockchain effectively prevents tampering, avoiding the theft and alteration of the charging log in plaintext, thus greatly improving the security of the charging log storage.

[0146] While specific embodiments of the present invention have been described above, those skilled in the art should understand that the specific embodiments described are merely illustrative and not intended to limit the scope of the invention. Equivalent modifications and variations made by those skilled in the art in accordance with the spirit of the invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for contactless charging of a charging station, characterized in that: Includes the following steps: Step S1: Create a face recognition model and set the loss function of the face recognition model; Step S2: Obtain a large number of historical face images, preprocess each of the historical face images to construct a dataset, train the face recognition model using the dataset, and deploy the trained face recognition model to the charging pile. Step S3: The charging pile uses dual-mode positioning technology of UWB and Bluetooth to locate the mobile terminal on the electric vehicle and obtain the location point. After compensating the location point with the RSSI of the mobile terminal, the distance between the charging pile and the vehicle is calculated based on the location point. The distance between the charging pile and the vehicle is then verified based on a preset distance threshold. Step S4: During the process of inserting the charging gun into the charging port of the electric vehicle, the charging pile collects real-time facial images, inputs the real-time facial images into the facial recognition model to obtain facial recognition results, and performs secondary verification based on the facial recognition results. Step S5: After the charging gun recognizes that the charging gun is plugged in, it interacts with the electric vehicle to perform a three-level verification. If the verification is successful, charging authorization is initiated, and then the electric vehicle is charged. Step S6: During the charging process of the charging pile to the electric vehicle, the charging strategy is dynamically adjusted until charging is completed and the charging authorization is closed. Step S7: The charging pile records the charging log and stores the charging log to the blockchain; Step S1 specifically involves: A face recognition model is created based on an image preprocessing module, a face feature extraction module, a feature comparison module, and a recognition module. The loss function of the face recognition model is set as Softmax Loss. The image preprocessing module, face feature extraction module, feature comparison module, and recognition module are connected sequentially. The image preprocessing module performs image normalization, size adjustment, and noise reduction on the input face image, and inputs the preprocessed face image into the face feature extraction module. The face feature extraction module extracts face features from the face image through convolutional layers, pooling layers, and activation functions, and inputs the face features into the feature comparison module. The feature comparison module calculates the cosine similarity between the face features and the face features registered in the database, and inputs the cosine similarity into the recognition module. The recognition module outputs the face recognition result based on the cosine similarity.

2. The contactless charging method for a charging pile as described in claim 1, characterized in that: Step S2 specifically involves: A large number of historical face images are acquired, and each of the historical face images is preprocessed, including at least cropping, resizing, illumination correction, grayscale conversion, normalization and noise reduction. After labeling the names of the people in each of the preprocessed historical face images, a dataset is constructed. The dataset is divided into a training set, a validation set, and a test set in a 7:2:1 ratio. The face recognition model is trained on the training set until the loss function's loss value is less than a preset loss threshold. During training, dynamic pruning and knowledge distillation techniques are used to compress the face recognition model. Next, the trained face recognition model is validated on the validation set to determine if the recognition accuracy is greater than a preset accuracy threshold. If not, the validation fails, and the dataset is expanded for continued training; if yes, the validation succeeds, and... The successfully validated face recognition model is tested using the test set to determine whether the recognition confidence level is greater than a preset confidence threshold. If not, the test fails, and the dataset is expanded to continue training; if yes, the test succeeds, training ends, and the trained face recognition model is deployed to the charging station.

3. The contactless charging method for a charging pile as described in claim 1, characterized in that: Step S4 specifically involves: During the process of inserting the charging gun into the charging port of the electric vehicle, the charging pile captures real-time facial images through a camera. After preprocessing the real-time facial images, they are input into a facial recognition model. The facial recognition model performs matching and recognition based on a pre-stored user management table to obtain facial recognition results. Secondary verification is then performed based on the facial recognition results. Step S5 specifically involves: After the charging gun recognizes that the charging gun is plugged in, it establishes a handshake communication with the electric vehicle, then interacts with the electric vehicle to obtain its operating status. Based on the operating status, it performs a three-level verification. If the verification passes, it initiates charging authorization and then charges the electric vehicle.

4. The contactless charging method for a charging pile as described in claim 1, characterized in that: Step S6 specifically involves: During the charging process of electric vehicles, the charging pile dynamically adjusts the charging strategy based on historical charging data, real-time electricity price and grid load until the preset charging cutoff conditions are met to complete the charging and close the charging authorization. Step S7 specifically involves: The charging station records charging logs in real time, including at least charging time, charging amount, electricity price, charging power, account, vehicle model, and vehicle serial number. The charging logs are encrypted using the SM4 algorithm to obtain a first layer of encrypted text. A 16-bit random string is generated and inserted into the beginning, middle, and end of the first layer of encrypted text to obtain a second layer of encrypted text. The second layer of encrypted text is then encrypted into a third layer of encrypted text using the IDEA algorithm. The numbers 0 and 0, 1 and 1, and 5 and 5 in the third layer of encrypted text are swapped to obtain a fourth layer of encrypted text. The fourth layer of encrypted text is then encrypted using the ECDSA algorithm to obtain an encrypted log, which is then stored on the blockchain.

5. A contactless charging system for charging piles, characterized in that: Includes the following modules: The face recognition model creation module is used to create a face recognition model and set the loss function of the face recognition model; The face recognition model deployment module is used to acquire a large number of historical face images, preprocess each of the historical face images to construct a dataset, train the face recognition model using the dataset, and deploy the trained face recognition model to the charging pile. The charging pile-vehicle distance calculation module is used to locate the mobile terminal on the electric vehicle using dual-mode positioning technology of UWB and Bluetooth, obtain the location point, compensate the location point by combining the RSSI of the mobile terminal, calculate the charging pile-vehicle distance based on the location point, and perform a first-level verification of the charging pile-vehicle distance based on a preset distance threshold. The face recognition module is used to collect real-time face images during the process of inserting the charging gun into the charging port of an electric vehicle, input the real-time face images into the face recognition model to obtain face recognition results, and perform secondary verification based on the face recognition results; The vehicle status verification module is used to interact with the electric vehicle to perform a three-level verification after the charging gun recognizes that the charging gun is plugged in. If the verification is successful, charging authorization is initiated, and then the electric vehicle is charged. The charging module is used to dynamically adjust the charging strategy during the charging process of electric vehicles by the charging pile until charging is completed and the charging authorization is closed. The charging log management module is used to record charging logs at the charging pile and store the charging logs on the blockchain; The face recognition model creation module is specifically used for: A face recognition model is created based on an image preprocessing module, a face feature extraction module, a feature comparison module, and a recognition module. The loss function of the face recognition model is set as Softmax Loss. The image preprocessing module, face feature extraction module, feature comparison module, and recognition module are connected sequentially. The image preprocessing module performs image normalization, size adjustment, and noise reduction on the input face image, and inputs the preprocessed face image into the face feature extraction module. The face feature extraction module extracts face features from the face image through convolutional layers, pooling layers, and activation functions, and inputs the face features into the feature comparison module. The feature comparison module calculates the cosine similarity between the face features and the face features registered in the database, and inputs the cosine similarity into the recognition module. The recognition module outputs the face recognition result based on the cosine similarity.

6. The contactless charging system for a charging pile as described in claim 5, characterized in that: The face recognition model deployment module is specifically used for: A large number of historical face images are acquired, and each of the historical face images is preprocessed, including at least cropping, resizing, illumination correction, grayscale conversion, normalization and noise reduction. After labeling the names of the people in each of the preprocessed historical face images, a dataset is constructed. The dataset is divided into a training set, a validation set, and a test set in a 7:2:1 ratio. The face recognition model is trained on the training set until the loss function's loss value is less than a preset loss threshold. During training, dynamic pruning and knowledge distillation techniques are used to compress the face recognition model. Next, the trained face recognition model is validated on the validation set to determine if the recognition accuracy is greater than a preset accuracy threshold. If not, the validation fails, and the dataset is expanded for continued training; if yes, the validation succeeds, and... The successfully validated face recognition model is tested using the test set to determine whether the recognition confidence level is greater than a preset confidence threshold. If not, the test fails, and the dataset is expanded to continue training; if yes, the test succeeds, training ends, and the trained face recognition model is deployed to the charging station.

7. The contactless charging system for a charging pile as described in claim 5, characterized in that: The face recognition module is specifically used for: During the process of inserting the charging gun into the charging port of the electric vehicle, the charging pile captures real-time facial images through a camera. After preprocessing the real-time facial images, they are input into a facial recognition model. The facial recognition model performs matching and recognition based on a pre-stored user management table to obtain facial recognition results. Secondary verification is then performed based on the facial recognition results. The vehicle status verification module is specifically used for: After the charging gun recognizes that the charging gun is plugged in, it establishes a handshake communication with the electric vehicle, then interacts with the electric vehicle to obtain its operating status. Based on the operating status, it performs a three-level verification. If the verification passes, it initiates charging authorization and then charges the electric vehicle.

8. The contactless charging system for a charging pile as described in claim 5, characterized in that: The charging module is specifically used for: During the charging process of electric vehicles, the charging pile dynamically adjusts the charging strategy based on historical charging data, real-time electricity price and grid load until the preset charging cutoff conditions are met to complete the charging and close the charging authorization. The charging log management module is specifically used for: The charging station records charging logs in real time, including at least charging time, charging amount, electricity price, charging power, account, vehicle model, and vehicle serial number. The charging logs are encrypted using the SM4 algorithm to obtain a first layer of encrypted text. A 16-bit random string is generated and inserted into the beginning, middle, and end of the first layer of encrypted text to obtain a second layer of encrypted text. The second layer of encrypted text is then encrypted into a third layer of encrypted text using the IDEA algorithm. The numbers 0 and 0, 1 and 1, and 5 and 5 in the third layer of encrypted text are swapped to obtain a fourth layer of encrypted text. The fourth layer of encrypted text is then encrypted using the ECDSA algorithm to obtain an encrypted log, which is then stored on the blockchain.

Images