| The security monitoring system plays a pivotal role in enabling the government to carry out effective social governance and maintain public security.While it offers convenience in terms of crime prevention and control,it also poses significant risks to citizens’ privacy and security.Of particular concern is the face information contained within these systems,as it constitutes personal information with strong social attributes that can be easily collected using biometric technology.The illegal collection,storage,use,and processing of this information can pose a serious threat to public safety,property,and personal dignity.To address this challenge,this thesis proposes a privacy protection and anti-tampering key technology for security surveillance image face data.(1)This thesis proposes a face detection algorithm based on the Swin Transformer network to address the low detection rates and missed detections of existing face detectors,particularly in the presence of face ornaments(such as helmets and masks)that obscure large portions of the face in video surveillance images.The proposed algorithm leverages the Swin Transformer as a feature extraction network to enhance feature extraction capabilities,and uses Decouple Head to improve the detection head,along with a new loss function design.Experimental analysis conducted on real security surveillance datasets demonstrates that the algorithm improves the detection rate of faces in such scenarios.(2)This thesis presents a novel scheme for face privacy protection and image storage that leverages blockchain technology.Specifically,the proposed scheme integrates blockchain with image processing technology to secure the entire process of video surveillance images,including acquisition,storage,and authorization.The scheme achieves this by securely storing separated face images while performing image face deprivation.The proposed solution not only safeguards citizens’ privacy by preventing the disclosure of private facial information but also stores face images in a secure and traceable manner for future access.The effectiveness of the scheme is demonstrated through empirical analysis conducted on real surveillance datasets and comparative analysis with other state-of-the-art techniques.(3)To mitigate the risks associated with the storage and use of face image data,this solution incorporates a blockchain tamper-proof module that incorporates a face fingerprint and face integrity verification process for face privacy images.The face fingerprint is derived by applying encryption and hashing algorithms to the face image data,resulting in a unique integrity credential that is securely stored using blockchain technology to ensure its authenticity.During face integrity verification,the face image digest contained within the face fingerprint can be used to quickly identify whether the corresponding face in the image has been tampered with,and the face location coordinates contained in the fingerprint can be used to pinpoint the location of the tampered face within the image. |
