Application of multi‐party computation and error correction with image enhancement and convolution neural networks based on cloud computing. Issue 6 (25th February 2023)

• Record Type: Journal Article
• Authors: Liao, Teh‐Lu; Peng, Chiau‐Yuan; Hou, Yi‐You
• Is Part Of: IET image processing. Volume 17:Issue 6(2023)
• Page Start: 1931
• Page End: 1950
• Publication Date: 2023-02-25
• Subjects: cloud computing -- convolutional neural nets -- image enhancement -- security of data; Image processing -- Periodicals; 621.36705
• DOI: 10.1049/ipr2.12767
• Languages: English
• ISSNs: 1751-9659

Abstract

Abstract: With the development of technology, people work hard on image processing and artificial intelligence which require a lot of computer resources. Instead of simply speeding up calculations with processors, people develop the technology of the Cloud with the Internet which performs high‐quality calculations and has quite a huge amount of storage. Thus, in this paper, the authors use a kind of encryption methods, Secure Multi‐Party Computation (SMPC), to protect the important information which is based on the absence of a trusted party and does not use specific keys or methods to maintain the confidentiality of data. Furthermore, the authors also use the error correction, Berlekamp–Welch (BW) algorithm, to double‐check the correction of the information. To apply the two algorithms to image enhancement and convolution neural networks (CNN), the authors will also discuss the homomorphism in all operations, including addition, multiplication, division etc., and find the best way to encrypt and decrypt the information. Finally, the authors will implement the system with cloud computing to decrease the consumption of computer resources and build a system that can do image enhancement and CNN. Abstract : After verifying the homomorphism and coming up with optimized ways, the authors build a system with image enhancement and convolutional neural networks on the cloud. Finally, with the system, the authors can achieve their goal that does the image enhancement and predicts the image with the encrypted image which will not only decrease the central processing unit (CPU) utilization of PC but also increase the security of information completely.