Level Set Stereo for Cooperative Grouping with Occlusion

TL;DR

This paper presents a level-set stereo method that enhances boundary localization in stereo images by explicitly modeling occlusions, leading to more accurate boundary detection in challenging scenes.

Contribution

It introduces an energy and level-set based optimizer that encodes occlusion geometry, improving boundary detection over previous methods.

Findings

More accurate boundary localization in stereo images.

Effective handling of occlusions in stereo matching.

Improved results on Middlebury and Falling Things datasets.

Abstract

Localizing stereo boundaries is difficult because matching cues are absent in the occluded regions that are adjacent to them. We introduce an energy and level-set optimizer that improves boundaries by encoding the essential geometry of occlusions: The spatial extent of an occlusion must equal the amplitude of the disparity jump that causes it. In a collection of figure-ground scenes from Middlebury and Falling Things stereo datasets, the model provides more accurate boundaries than previous occlusion-handling techniques.