Shape Reconstruction Based on Polarization and Coarse Depth Images

doi:10.13438/j.cnki.jdzk.2022.04.005

Abstract

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

ZHOU Mengyuan, YANG Fenlin. Shape Reconstruction Based on Polarization and Coarse Depth Images[J]. Journal of Jishou University(Natural Sciences Edition), 2022, 43(4): 20-26.

References

[1] RAHMANN STEFAN,CANTERAKI NIKOS.Reconstruction of Specular Surfaces Using Polarization Imaging[C]∥Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.CVPR 2001.Kauai,HI,USA:IEEE,2001:149-155.
[2] NGO THANH TRUNG,NAGAHARA HAJIME,TANIGUCHI RIN-ICHIRO.Shape and Light Directions from Shading and Polarization[C]∥2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).Boston,MA,USA:IEEE,2015:2310-2318.
[3] IZADI SHAHRAM,KIM DAVID,HILLIGES OTMAR,et al.KinectFusion:Real-Time 3D Reconstruction and Interaction Using a Moving Depth Camera[C]∥PIERCE JEFF.Proceedings of the 24th Annual ACM Symposium on User Interface Software and Technology.New York,NY,USA:Association for Computing Machinery,2011:559-568.
[4] CUI Zhaopeng,GU Jinwei,SHI Boxin,et al.Polarimetric Multi-View Stereo[C]∥ 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).Honolulu,HI,USA:IEEE,2017:1558-1567.
[5] KADAMBI ACHUTA,TAAMAZYAN VAGE,SHI Boxin,et al.Depth Sensing Using Geometrically Constrained Polarization Normals[J].International Journal of Computer Vision,2017,125(1):34-51.
[6] 杨锦发,晏磊,赵红颖,等.融合粗糙深度信息的低纹理物体偏振三维重建[J].红外与毫米波学报,2019,38(6):819-827.
[7] 张瑞华,施柏鑫,杨锦发,等.基于视差角和天顶角优化的偏振多视角三维重建[J].红外与毫米波学报,2021,40(1):133-142.
[8] FLEISHMAN SHACHAR,DRORI IDDO,COHEN-OR DANIEL.Bilateral Mesh Denoising [J].ACM Transactions on Graphics,2003(3):950-953.
[9] 公羽,吕志鹏.基于双边滤波的点云数据预处理方法[J].测绘与空间地理信息,2016,39(1):125-127.
[10] KLASING KLAAS,ALTHOFF DANIEL,WOLLHERR DIRK,et al.Comparison of Surface Normal Estimation Methods for Range Sensing Applications[C]∥ 2009 IEEE International Conference on Robotics and Automation.Kobe,Japan:IEEE,2009:3206-3211.
[11] ATKINSON GARY A,HANCOCK EDWIN R.Shape Estimation Using Polarization and Shading from Two Views[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2007,29(11):2001-2017.
[12] NEHAB DIEGO,RUSINKIEWICZ SZYMON,DAVIS JAMES,et al.Efficiently Combining Positions and Normals for Precise 3D Geometry[J].ACM Transactions on Graphics (TOG),2005,24(3):536-543.
[13] YU Ye,SMITH WILLIAM A P.Depth Estimation Meets Inverse Rendering for Single Image Novel View Synthesis[C]∥ SMITH WILLIAM A P.European Conference on Visual Media Production.New York,NY,USA:Association for Computing Machinery,2019:1-7.
[14] FRANKOT ROBERT,CHELLAPPA RAMA.A Method for Enforcing Integrability in Shape from Shading Algorithms[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1988,10(4):439-451.