Secure Image LBP Feature Extraction in Cloud-Based Smart Campus

Abstract

The smart campus can monitor students in real time by analyzing students' images, but a large number of images bring an unbearable burden to the smart campus. The convenience of cloud computing has attracted smart campus to outsource their huge amount of data to cloud servers. Although the outsourcing of data can reduce the computational and storage burden on smart campus, the privacy preserving becomes the biggest concern. This issue has attracted many researchers to study the protection of outsourced multimedia data. In this paper, we propose an effective and practical privacy-preserving computation outsourcing protocol for the local binary pattern (LBP) feature over huge encrypted images. The image owner uploads the encrypted version of images to the cloud. The cloud server takes the responsibility of extracting the LBP features from encrypted images for various applications. In the encryption process, an image is divided into non-overlapping blocks at first, and the blocks are shuffled to protect the image content. Next, all the non-center pixels in each block are shuffled. Finally, the pixels are encrypted by splitting the original image data randomly. When such an encrypted image is received, the cloud servers can calculate the LBP features by secure multiparty computation. The extracted features can be applied to many applications, such as texture classification, image retrieval, face recognition, and so on.