【摘 要】
：
We extend the existing variational model for image inpainting and propose new coupled and decoupled algorithms.In the derivation of coupled algorithm,we use
【出 处】
：
Computational Biomedical Imaging Workshop(2015计算生物医学成像研讨会)
论文部分内容阅读
We extend the existing variational model for image inpainting and propose new coupled and decoupled algorithms.In the derivation of coupled algorithm,we use operator splitting and quadratic penalty technique to get a new approximate problem of the basic model.By alternating minimization method,the approximate problem can be decomposed as several relatively simple subproblems with closed-form solutions.However,the coupled algorithm is not efficient when some adaptive regularization operators are used such as learned BM3D.To overcome this drawback,we propose the decoupled algorithm in which the original problem is decoupled into several independent parts: denoising and linear combinations.Therefore,we can take use of any existing denoising method in the denoising step.We consider three choices for regularization operator in our experiments which are gradient operator,tight framelet transform and learned BM3D frame.The numerical experiments and comparisons on various image inpainting tasks demonstrate that the proposed method is promising.
其他文献
In this talk,I will propose a convex model for simultaneous image restoration and segmentation based on global TV.We adopt the widely used ADMM to solve the
Parameter mapping is a powerful noninvasive technique providing quantitative biological information of the inherent tissue properties.In this talk,we will i
To improve the low-dose CT(LDCT)imaging quality,we proposed a new processing approach based on the morphological component analysis(MCA),with the integratio
I will start with a brief review of existing models(both classical and recent)for solving ill-posed linear inverse problems,that include wavelet frame based
In this talk,we focus on the challenging problem of removing the spatially varying out-of-focus blur from a single natural image.We first propose an effecti
Images by reflection imaging can be well approximated by piecewise constant functions.Thus total variation is an essential tool to regularize this kind of i
In some practical applications,restricted by scanning environment or scanned object,the projection data obtained from CT imaging system is incomplete,which
In this paper,aboundary-type meshless numerical scheme,based on the singular boundary method(SBM),is proposed to efficiently analyze the two-dimensional Sig
Based on the recently developed MFS-based particular solution for two-dimensional problems in isotropic linear thermoelastic materials [1],we investigate th
Spectral data across energy bins are highly correlated but the signal to noise ratio is generally poor for each energy bin due to the narrow bin width and r