Edge-guided second-order total generalized variation for Gaussian noise removal from depth map

Abstract

Total generalized variation models have recently demonstrated high-quality denoising capacity for single image. In this paper, we present an accurate denoising method for depth map. Our method uses a weighted second-order total generalized variational model for Gaussian noise removal. By fusing an edge indicator function into the regularization term of the second-order total generalized variational model to guide the diffusion of gradients, our method aims to use the first or second derivative to enhance the intensity of the diffusion tensor. We use the first-order primal–dual algorithm to minimize the proposed energy function and achieve high-quality denoising and edge preserving result for depth maps with high -intensity noise. Extensive quantitative and qualitative evaluations in comparison to bench-mark datasets show that the proposed method provides significant higher accuracy and visual improvements than many state-of-the-art denoising algorithms.

Denoising on Textured Image Using Total Generalized Variation With Overlapping Group Sparsity Based on Fast Split Bregman Method

“…In comparison to the L1 operator, the Lp contraction operator introduces an additional degree of freedom, allowing for a more accurate representation of sparse gradient information in the image. However, this enhancement comes at the cost of increased computational complexity [23], [24], [25]. Ban et al [26] introduced a non-local self-similarity in the transform domain as a prior TGV information, incorporating a multi-directional TGV regularization constraint calculated within the eight-neighborhood space to safeguard the structural characteristics of the image.…”

Enhancing low-rank tensor completion via first-order and second-order total variation regularizations

“…Both theoretical and experimental researches show that the TV norm transforms smooth signal into piecewise constants, i.e. so-called staircase effects, which can lead to false edges that do not exist and fail to satisfy the visual evaluation [12,17]. Therefore, inspired by second-order TV regularization in low-rank matrix recovery [34,39], some works have considered the second-order regularization in LRTC.…”