A novel privacy-preserving outsourcing computation scheme for Canny edge detection

With the advancement of cloud computing technology, cloud servers are utilized to process large-scale data, especially multimedia data. However, concerns about leakage of private information prevent cloud computing from being further popularized. Thus, privacy-preserving computation for multimedia data is becoming increasingly important as a research hotspot. Edge detection plays an important role in image processing and computer vision. Different from previous researches on privacy-preserving computation, privacy-preserving edge detection faces new problems such as how to encrypt and represent edges. In this paper, we propose a privacy-preserving computation scheme for Canny edge detection. We first give an overview of our scheme, which involves one client and three cloud servers. Then, three key building blocks in the proposed scheme are put forward: pixel permutation; secure comparison and multiplication protocols; secure edge representation. Based on these building blocks, our scheme is carefully designed and constructed step by step. Furthermore, we analyze the correctness and the security of our scheme in detail. Finally, comparative experiments show that our scheme can maintain the quality of edge detection while meeting security requirements.