Research On Secure Outsourced Denoising Schemes In Encrypted Remote Sensing Images Based On Double-Cipher Policy

With the rapid development of remote sensing technology,remote sensing images can provide more and more abundant surface-feature information and have more advantages including more access to acquiring information means,short period of detection,characteristic of multi-bands,real-time and spatiality.They play an indispensable role in the disaster forecast and hazard assessment,ecological environment monitoring,air quality monitoring,weather forecast,military and national defense,and many other areas,which also becomes an important ways in globally data-dynamic observation.However,remote sensing images during the process of acquiring,transmitting,receiving and exporting,often tend to be disturbed by noise,which makes remote sensing images appear the fuzzy edges of the image texture details,decrease the quality of the images and amplifies the difficulty in the post-processing,analysis and application of images and affects the value of data.Therefore,remote sensing images must be pre-processed by denoising.The processing on huge number of the remote sensing images on the ordinary computers leads to small storage,low processing speed and poor concurrency,and so on,has been unable to meet the needs of massive remote sensing image processing.Outsourced computing mode can provide a huge-amount storage capacity,flexible computing ability,powerful concurrent performance,and simultaneously shorten the execution time of computing tasks.However,cloud platforms are not entirely reliable,and data stored on the cloud platform is often lost or stolen due to the lack of regulation and supervision,leading to an incident of important data leakagean.Remote sensing images,as an important strategic resources of the country,need to prevent the leakage of sensitive information.In order to safeguard the security of remote sensing data,we need to encrypt those data stored on the cloud platform.The data processed based on the traditional encryption methods generally cannot be further utilized.Therefore,it is urgent to implement the denoising in the encrypted remote sensing images on the cloud platform.According to the above analysis,this paper proposes an outsourced denoising scheme based on double-cipher policy to implement non-local means(NLM)denoising in encrypted remote sensing images,and the main contributions are following three aspects:(1)This paper presents a scheme for the secure outsourcing denosing of remote sensing images.In the scheme,the ciphertext from Johnson-Lindenstrauss transform is used to weight calculation as well as the plaintext and the ciphertext from Paillier homomorphic encryption is used to nonlocal means denoising by the linear calculation of ciphertext.(2)This paper also proposes an improved scheme for the outsourcing denosing of remote sensing images.In order to enhance the security of the secure outsourcing denosing scheme,the method of image obfuscation based on pseudo-random sequence is used to randomly permutate the pixels and the pixel blocks.In order to improve the efficiency of the secure outsourcing denosing scheme,improved Paillier encryption method and non-local means denoising algorithm substitute for the original corresponding methods,respectively.(3)This paper furthuer proposes a scheme for security outsourcd fusion denoising of multiple remote sensing images.The scheme uses the correlation of multiple images in time and space domains,find more pixels in higher similar degrees in the process of non-local denoising,better weaken the noise in the image,and implement outsourced fusion denoising in encrypted remote sensing images based dynamic-filtering parameters.The authors cut out several 512×512 pixels images from Landsat 8 remote sensing images and use the seven bands(Coastal,Blue,Green,Red,and NIR,SWIR1 and SWIR2)whose spatial resolution is 30 meters as the experimental images.The experimental results demonstrate that the scheme of outsourcing denoising can effectively guarantee the security of the remote sensing images,obtain better denoising effect with different sizes of noise,and increase the efficiency of non-local mean denoising processing.In this paper,the proposed method,outsourcing fusion denoising,has apparently superior effects in the respect of denoising than existing methods of cipher-text denoising and single-frame denoising,and has good denoising effects for different sizes of noise.As the numbers of frames are improving,the effect of denoising becomes significantly better.