Underwater multi-spectral photometric stereo reconstruction from a single RGBD image

Traditional theories and methods in 3D reconstruction were all proposed with implicit assumption of air environment. However, underwater environment is different in many aspects. The absorption and scattering effects caused by the suspended particles in the water attenuate the image signal, which disqualifies the traditional reconstruction algorithms. In this paper, we propose a novel method to reconstruct high-resolution 3D surface of underwater objects by fusing depth and multispectral photometric stereo based on a single RGBD image. First, we use depth sensor to obtain the object's RGB image with depth information. Then, we remove backscatter by fitting a binary quadratic function and apply simple linear iterative clustering (SLIC) Superpixels to segment the RGB image. Based on these superpixels, we use multispectral photometric stereo(PS) to calculate the objects' surface normal. Since initial depth data contains noise and not enough high-frequency detail, surface normal from PS are fused with depth data for high-resolution 3D reconstruction. Preliminary experimental results demonstrate that our method achieves more accurate estimates of surface normal and 3D reconstruction from a single frame.