High-level visual features for underwater place recognition
Jie Li · R. Eustice · M. Johnson-Roberson
TL;DR
Matching salient image patches instead of points lets underwater robots recognize places across years despite lighting changes and biofouling.
BRIEF
Underwater robots that revisit a site months or years later struggle to recognize where they are. Lighting changes, biofouling and growth alter how surfaces look, and few distinctive points exist to match. This work replaces point features with salient image patches, found through segmentation and local intensity contrast and matched using HOG descriptors. Patches capture shape and structure, which tend to persist across years better than fine detail. The method is tested on multi-year ship hull inspection data from a hovering underwater vehicle and compared with traditional approaches.
Abstract
This paper reports on a method to perform robust visual relocalization between temporally separated sets of underwater images gathered by a robot. The place recognition and relocalization problem is more challenging in the underwater environment mainly due to three factors: 1) changes in illumination; 2) long-term changes in the visual appearance of features because of phenomena like biofouling on man-made structures and growth or movement in natural features; and 3) low density of visually salient features for image matching. To address these challenges, a patch-based feature matching approach is proposed, which uses image segmentation and local intensity contrast to locate salient patches and HOG description to make correspondences between patches. Compared to traditional point-based features that are sensitive to dramatic appearance changes underwater, patch-based features are able to encode higher level information such as shape or structure which tends to persist across years in underwater environments. The algorithm is evaluated on real data, from multiple years, collected by a Hovering Autonomous Underwater Vehicle for ship hull inspection. Results in relocalization performance across missions from different years are compared to other traditional methods.