Authors: Ping Gong; Guohua Li; Jian Li
Addresses: School of Information and Electronic Engineering, Wuzhou University, Wuzhou Guangxi 543002, China; Guangxi Colleges and Universities Key Laboratory of Image Processing and Intelligence Information System, Wuzhou University, Wuzhou Guangxi 543002, China ' School of Information and Electronic Engineering, Wuzhou University, Wuzhou Guangxi 543002, China ' School of Information and Electronic Engineering, Wuzhou University, Wuzhou Guangxi 543002, China
Abstract: This paper proposed an approach combines the advantages of L1 norm and TV norm by combining L1 norm and TV norm to solve the image reconstruction problems. And the proposed approach reconstructs an image from the measured values by using the modified conjugate gradient algorithm for the purpose of improving the efficiency of image reconstruction. The objective function is constructed using the approximate function based on the L1 norm and TV norm. A sparse transformation is applied to the original image first. The random Gaussian matrix is then employed to perform a compressive measurement on the 2-D sparse signal. To reconstruct the image a regularised reconstruction model is established through the approximate norm that combines L1 norm and TV norm and the gradient of the approximate norm is computed. The simulation results demonstrate the ability of the proposed method to reconstruct images more effectively and produce better results in terms of objective indicators such as PSNR and SSIM than classical methods.
Keywords: compressive sensing; L1 norm; total variation; modified conjugate gradient algorithm; image reconstruction.
International Journal of Computing Science and Mathematics, 2020 Vol.11 No.1, pp.93 - 105
Received: 16 Aug 2017
Accepted: 25 Sep 2017
Published online: 27 Feb 2020 *