电子、通信与自动控制

基于仿射投影的光线空间插值方法

展开
  • 华南理工大学 电子与信息学院, 广东 广州 510640
孙季丰(1962-),男,教授,博士生导师,主要从事通信与信息系统、图像与视频处理研究.

收稿日期: 2012-06-08

  修回日期: 2012-10-24

  网络出版日期: 2013-01-05

基金资助

广东省自然科学基金资助项目(9151064101000037)

Ray-Space Interpolation Based on Affine Projection

Expand
  • School of Electronic and Information Engineering, South China University of Technology, Guangzhou 510640, Guangdong, China
孙季丰(1962-),男,教授,博士生导师,主要从事通信与信息系统、图像与视频处理研究.

Received date: 2012-06-08

  Revised date: 2012-10-24

  Online published: 2013-01-05

Supported by

广东省自然科学基金资助项目(9151064101000037)

摘要

为有效处理不同视点图像中对应区域的不同形变和噪声污染,提高所绘制的视点图像的质量和速度,提出了一种基于仿射投影的光线空间插值方法. 该方法首先利用尺度不变特征变换进行特征检测和图像配准,计算出相应图像对之间的仿射变换矩阵,然后利用仿射变换投影进行光线空间插值,并在小波变换域绘制出虚拟视点图像. 实验结果表明,采用文中方法绘制的虚拟视点图像的峰值信噪比和计算时间均优于传统的块匹配插值和视差域滤波插值方法.

本文引用格式

孙季丰 吴军政 . 基于仿射投影的光线空间插值方法[J]. 华南理工大学学报(自然科学版), 2013 , 41(2) : 1 -6 . DOI: 10.3969/j.issn.1000-565X.2013.02.001

Abstract

In order to effectively deal with the shape variation and noise pollution of the corresponding area in diffe-rent views and, thereby, improve the quality and speed of viewpoint image synthesis, a ray-space interpolationmethod based on affine projection is proposed. In this method, first, the scale-invariant feature transform is appliedto the feature detection and image matching for determining the affine transformation matrix among multi-viewpointimages. Then, ray-space interpolation is implemented via the projection through affine transformation matrix. Finally,a virtual viewpoint image is synthesized in the wavelet transform domain. Experimental results show that the pro-posed method is superior to the traditional block-matching interpolation and DDFI (Disparity Domain Filtering In-terpolation) methods in terms of PSNR and computation time of the virtual viewpoint image.

文章导航

/