Polarimetric PatchMatch Multi-View Stereo
This work addresses a domain-specific challenge in computer vision for 3D reconstruction, particularly benefiting applications requiring detailed models of texture-less objects, and is incremental as it builds upon existing PatchMatch MVS with a novel cue.
The paper tackled the problem of improving 3D reconstruction accuracy and completeness in multi-view stereo, especially for texture-less surfaces, by introducing polarimetric consistency into PatchMatch MVS, resulting in enhanced performance compared to state-of-the-art methods.
PatchMatch Multi-View Stereo (PatchMatch MVS) is one of the popular MVS approaches, owing to its balanced accuracy and efficiency. In this paper, we propose Polarimetric PatchMatch multi-view Stereo (PolarPMS), which is the first method exploiting polarization cues to PatchMatch MVS. The key of PatchMatch MVS is to generate depth and normal hypotheses, which form local 3D planes and slanted stereo matching windows, and efficiently search for the best hypothesis based on the consistency among multi-view images. In addition to standard photometric consistency, our PolarPMS evaluates polarimetric consistency to assess the validness of a depth and normal hypothesis, motivated by the physical property that the polarimetric information is related to the object's surface normal. Experimental results demonstrate that our PolarPMS can improve the accuracy and the completeness of reconstructed 3D models, especially for texture-less surfaces, compared with state-of-the-art PatchMatch MVS methods.