Journal of Jishou University(Natural Sciences Edition) ›› 2022, Vol. 43 ›› Issue (4): 20-26.DOI: 10.13438/j.cnki.jdzk.2022.04.005

• Computer • Previous Articles     Next Articles

Shape Reconstruction Based on Polarization and Coarse Depth Images

ZHOU Mengyuan,YANG Fenlin   

  1. (School of Mathematics and Statistics,Jishou University,Jishou 416000,Hunan China)
  • Online:2022-07-25 Published:2022-10-28

Abstract: Firstly,the coarse depth image is smoothed by bilateral filtering,which can remove noise well while keeping edge details.Then,the depth information is taken as a prior condition to correct the ambiguity of polarization normals by using a self-defined threshold segmentation method.Finally,a penalty function method with depth constraint as the fitting term and polarization normal constraint as the smooth term is constructed,and the least square method is used to calculate the fusion reconstruction.Four polarization images with different polarization angles and a coarse depth map are selected for experiments.The experimental results show that compared with single mode,the fusion of depth and polarization can recover better surface details and achieve better geometric shape effect.

Key words: polarization, coarse depth map, bilateral filtering, multimodal fusion, threshold segmentation

WeChat e-book chaoxing Mobile QQ