Normal reconstruction from specularity in the endoscopic setting

TL;DR

This paper introduces a method to reconstruct surface normals in endoscopic images by analyzing specular isophotes, which appear as nested ellipses, enabling normal estimation on moist tissue surfaces.

Contribution

The paper presents a novel technique to estimate surface normals from specular reflections in endoscopic images, validated on simulated and real medical data.

Findings

Specular isophotes form concentric circles on scene plane

Ellipses in images can be used to estimate surface normals

Method applied successfully to laparoscopic and colonoscopic images

Abstract

We show that for a plane imaged by an endoscope the specular isophotes are concentric circles on the scene plane, which appear as nested ellipses in the image. We show that these ellipses can be detected and used to estimate the plane's normal direction, forming a normal reconstruction method, which we validate on simulated data. In practice, the anatomical surfaces visible in endoscopic images are locally planar. We use our method to show that the surface normal can thus be reconstructed for each of the numerous specularities typically visible on moist tissues. We show results on laparoscopic and colonoscopic images.