Double-blind privacy preserving face retrieval method and system

Abstract

The invention relates to a double-blind privacy protection face retrieval method and system. The method comprises the following steps that: 1) a client end detects an image containing a suspected missing person or an image containing a real missing person to obtain a face image; 2) the client carries out feature extraction on the detected face image to obtain a face representation vector; 3) thatclient end encrypts the face image and the extracted face representation vector and upload the encrypted face image and the extracted face representation vector to the cloud server; The cloud server matches the face image information of the suspected missing person with the face image information of the real missing person in the encrypted domain by using the face representation vector, obtains the matching result, and decrypts the adult face image by the matching result; 4) The client end receives the face image decrypted according to the matching result from the cloud server, and confirms whether the face image is a real missing person. The invention has good privacy protection, high precision and processing speed, and can provide an effective recovery scheme for missing persons, especially missing children.

Description

technical field

[0001] The invention belongs to the field of artificial intelligence and information security, and in particular relates to a child face retrieval method and system under double-blind privacy protection, and is especially suitable for the field of child face retrieval. Background technique

[0002] According to statistics, there are about 460,000, 113,000, and 100,000 children missing in the United States, Britain, and Germany every day around the world. Based on the fact that face is a basic attribute for distinguishing individuals, in order to retrieve lost children, some effective means have been developed, for example, through the Center of Missing & Exploited Children (CME), missing children Parents can publish the photo information including the time and location of the disappearance and the real missing child. If someone sees the missing child, they can tell where and when it was found, so as to help find the missing child.

[0003] In recent years, w...

Examples

Embodiment Construction

[0044] In order to make the above solutions and beneficial effects of the present invention more comprehensible, the following is a detailed description through examples and accompanying drawings.

[0045]This embodiment takes children's face retrieval as an example, and provides a double-blind privacy-protected child's face retrieval method and a system for implementing the method. The method includes the following steps:

[0046] Use a pre-trained face detector to perform face detection on the photos of suspicious missing children taken by SNH, and use a pre-trained face recognizer to extract multi-attribute feature vectors from the detected children's faces;

[0047] Use the extracted multi-attribute feature vector and the public parameters provided by CME to generate a pair of public keys PK F and PK I . The public parameter can be a vector with all 1s, or other suitable forms.

[0048] Using PK F Encrypt the matching token F to form a ciphertext token C F ;F is a bi...

technical field

[0001] The invention belongs to the field of artificial intelligence and information security, and in particular relates to a child face retrieval method and system under double-blind privacy protection, and is especially suitable for the field of child face retrieval. Background technique

[0002] According to statistics, there are about 460,000, 113,000, and 100,000 children missing in the United States, Britain, and Germany every day around the world. Based on the fact that face is a basic attribute for distinguishing individuals, in order to retrieve lost children, some effective means have been developed, for example, through the Center of Missing & Exploited Children (CME), missing children Parents can publish the photo information including the time and location of the disappearance and the real missing child. If someone sees the missing child, they can tell where and when it was found, so as to help find the missing child.

[0003] In recent years, w...

Examples

Embodiment Construction

[0044] In order to make the above solutions and beneficial effects of the present invention more comprehensible, the following is a detailed description through examples and accompanying drawings.

[0045]This embodiment takes children's face retrieval as an example, and provides a double-blind privacy-protected child's face retrieval method and a system for implementing the method. The method includes the following steps:

[0046] Use a pre-trained face detector to perform face detection on the photos of suspicious missing children taken by SNH, and use a pre-trained face recognizer to extract multi-attribute feature vectors from the detected children's faces;

[0047] Use the extracted multi-attribute feature vector and the public parameters provided by CME to generate a pair of public keys PK F and PK I . The public parameter can be a vector with all 1s, or other suitable forms.

[0048] Using PK F Encrypt the matching token F to form a ciphertext token C F ;F is a bi...

Images