Error-Tolerant Semi-Automatic 2D-to-3D Conversion via L<sub>1</sub> Optimization

YUAN Hong-xing; AN Peng; WU Shao-qun; ZHENG You

doi:10.3969/j.issn.0372-2112.2018.02.026

您当前的位置：

首页 >

文章列表页 >

Error-Tolerant Semi-Automatic 2D-to-3D Conversion via L₁ Optimization

更新时间：2025-07-16

- Error-Tolerant Semi-Automatic 2D-to-3D Conversion via L₁ Optimization
- Acta Electronica Sinica Vol. 46, Issue 2, Pages: 447-455(2018)
- 作者机构：
  
  宁波工程学院电子与信息工程学院,浙江,宁波,315211
- 作者简介：
- 基金信息：
- DOI：10.3969/j.issn.0372-2112.2018.02.026
  CLC： TN911.73
- Published Online：25 February 2018，
  
  Published：2018
- 稿件说明：
移动端阅览
YUAN Hong-xing, AN Peng, WU Shao-qun, et al. Error-Tolerant Semi-Automatic 2D-to-3D Conversion via L₁ Optimization[J]. Acta Electronica Sinica, 2018, 46(2): 447-455.
DOI：

YUAN Hong-xing, AN Peng, WU Shao-qun, et al. Error-Tolerant Semi-Automatic 2D-to-3D Conversion via L₁ Optimization[J]. Acta Electronica Sinica, 2018, 46(2): 447-455. DOI： 10.3969/j.issn.0372-2112.2018.02.026.

摘要

半自动2D转3D的关键是将用户分配的稀疏深度转换为稠密深度.现有方法没有充分考虑纹理图像和深度图之间的结构差异，以及2D转3D对用户误标注的容错性.针对上述问题，借助

范数对异常数据的抵制，在一个统一框架下实现结构相关具有容错能力的稀疏深度稠密插值.首先，利用

范数表示估计深度和用户分配深度在标注位置的差异，建立数据项；其次，根据特征的相似性用

范数计算局部相邻像素点之间的深度差异，建立局部正则项；再次，对图像进行超像素分割，根据不同超像素内代表性像素点之间深度差异的

测度，建立全局正则项；最后，用上述数据项和正则项构建能量函数，并通过分裂Bregman算法予以求解.无误差和有误差情况下的实验结果表明，与边缘保持的最优化插值、随机游走、混合图割与随机游走、软分割约束的最优化插值和非局部化随机游走相比，本文估计深度图绘制的虚拟视点图像空洞和伪影损伤更小.在误操作情况下，本文比上述方法PSNR改善了0.9dB以上，且在视觉上屏蔽了用户误操作的影响.

Abstract

Sparse-to-dense depth conversion is an important task in semi-automatic 2D-to-3D conversion. Existing methods do not handle structural difference between texture image and depth map

and the error-tolerance of 2D-to-3D is not considered. Inspired by compressive sensing studies

we address these problems in an optimization framework via

norm. First

data term is built with

norm to measure the fidelity between estimated depth and user ass

igned depth. Second

local regularized term is defined by using feature weighted

norm to measure difference between local neighboring pixels. Third

super-pixels are generated from input image and global regularized term is introduced by using feature weighted

norm to measure difference between representative pixels from these super-pixels. Then

the energy function for sparse-to-dense depth conversion is defined based on the data term

local regularized term and global regularized term. The split Bregman algorithm is used to solve the energy. Experimental comparisons with optimization based interpolation

random-walks

hybrid graph-cuts and random-walks

soft segmentation constrained interpolation and nonlocal random-walks show that our method demonstrates significant advantages over hole and ghosting artifacts for viewpoint synthesis. The PSNR is improved by more than 0.9 dB compared with these methods when user assigns error depth.

关键词

Keywords

references

Views

281

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Semi-Automatic 2D-to-3D Conversion Using Soft Segmentation Constrained Edge-Aware Interpolation

An Optimal Transformation of Basic Probability Assignment to Probability

Autonomous Learning Path Selection Scheme for Concurrent Multipath Transfer

Approach of Traffic Matrix Estimation in Large-scale IP Backbone Networks

Study of Dual Linear Programming Neural Networks and Selection of Its Learning Rate

Related Author

CHENG Pei-hong

AN Peng

YUAN Hong-xing

WU Shao-qun

DENG Yong

HAN De-qiang

XU Pei-da

ZHANG Si-dong

Related Institution

School of Electronics and Information Engineering, Ningbo University of Technology

College of Computer and Information SciencesSouthwest UniversityChongqing 400715China

Institute of Integrated AutomationXi’an Jiaotong UniversityXi’anShaanxi 710049China

School of ElectronicInformation and Electrical EngineeringShanghai Jiaotong UniversityShanghai 200240China

College of Computer and Information Sciences,Southwest University

⁰

Error-Tolerant Semi-Automatic 2D-to-3D Conversion via L1 Optimization

DOI：10.3969/j.issn.0372-2112.2018.02.026

摘要

Abstract

关键词

Keywords

references

Error-Tolerant Semi-Automatic 2D-to-3D Conversion via L₁ Optimization