Research On LBP Feature Extraction Scheme In Encrypted Images

In the era of Industrial Internet of Things(IIoT),huge amounts of image data is generated,which imposes a great storage burden on data owners.The rapid development of cloud computing has attracted data owners to store data on cloud servers for cost savings and efficient applications.However,the IIoT data is always of great value.The direct outsourcing of such data can leak the important information of data owners.A straightforward solution is to encrypt the data with standard encryption tools before outsourcing.However,this will cause difficulties in extracting image features.This paper focuses on extracting Local Binary Pattern(LBP)from encrypted images in cloud servers.The images are stored on the cloud server in encrypted form,and the LBP feature can be directly extracted from the encrypted images for applications.In this paper,we are committed to researching effective and practical privacy protection LBP feature extraction outsourcing algorithms.The image owner uploads the encrypted image database to the cloud server.The cloud server is responsible for extracting LBP features from the encrypted images of various applications.Research on LBP feature extraction algorithm based on order-preserving encryption(referred to as SC-LBP).This scheme protects the privacy of the images by dividing the image into blocks,scrambling the blocks,scrambling the pixels in the image blocks and order-preserving encryption.Only one cloud server is needed in this scheme,which extracts LBP features from non-overlapping blocks of the encrypted images.Research on LBP feature extraction algorithm based on secure multi-party computation(referred to as MC-LBP).This scheme protects the privacy of the images by dividing the image into blocks,scrambling the blocks,scrambling the pixels in the image blocks and image segmentation.Ttwo cloud entities are needed in this scheme.The primary server and the secondary server exchange secure interactions to extract the LBP features from the non-overlapping blocks.The LBP features are extracted from the non-overlapping block by secure interactions between the primary server and the secondary server.Through security analysis,the security of SC-LBP scheme is better than MC-LBP scheme,but they are still secure.The experimental results show that the effects of SC-LBP and MC-LBP schemes are basically the same.MC-LBP is better than SC-LBP in time complexity,but both of them are effective.