Research On JPEG Image Decoding In Homomorphic Encryption Domain

In the cloud computing environment,user data is transferred to the cloud for processing,and users lose direct control over the data.Homomorphic Encryption supports computation on ciphertext,and can realize cloud data processing on the premise of protecting user data security.Existing Homomorphic field JPEG image ciphertext compression scheme JDHE(JPEG Decompression in the Homomorphic Encryption Domain)basically outputs only one coefficient per cycle.There is still room for improvement in computing efficiency.New scheme FJDHE(Faster JPEG Decompression in the Homomencryption Domain)can parse and output multiple coefficients per round cycle,reducing the number of iteration cycles.Moreover,in each round iteration process,there is no need to insert new codewords into the Huffman table,and no need to insert the Huffman codewords into the cryptography stream for round-by-round iterative updating,thus improving the computing efficiency.In addition,the new scheme further proposed performance optimization measures.1)In order to reduce the frequency of time-consuming noise reduction operation,multi-ciphertext continuous multiplication algorithm is optimized by using tree operation structure to select ciphertext with the same or similar noise level for operation to reduce the loss of ciphertext noise level.2)In order to increase parallelism,ciphertext packaging technology is used to support single-instruction multi-data flow operation,and parallel operations such as update code stream and update coefficient are not dependent on each other.In addition,it can be combined with data characteristics and application scenarios to flexibly configure different parameters,select picture quality and time complexity to achieve the optimal performance.The experimental results first show the improved effect of optimizing ciphertext continuous multiplication algorithm,and then verify the effectiveness of Homomorphic Encryption domain JPEG decomcompression algorithm of the new scheme FJDHE,compared with JDHE improves the computational efficiency.Fully decoding a 128x128 JPEG grayscale image reduces the usage time by 32.93% to 56.71%.