A Discrete Bayes Filter for visual loop-closing in image sequences of coral reef explorations taken by humans and AUVs

摘要

The recognition of places by using visual information in underwater environments is important when performing autonomous robotic exploration of the same area at different periods of time. It helps the robot to know its location and take decisions accordingly. However, vision-based recognition of underwater places can be a very challenging task due to the inherent properties of this kind of places such as: color distortion, poor visibility, high perceptual aliasing and dynamic lighting conditions. In this work we focus on the problem of finding whether two images from an sequence belongs to the same place or not. For the case of few images, this can be a trivial task, however when several images have to be taken into account, comparing one by one becomes intractable. In this paper we have used a Discrete Bayes Filter for detecting loop closure events using visual information, that is when two images are from the same place. We have tested the proposed approach using sequences of images obtained from explorations of coral reefs taken by divers and autonomous underwater vehicles. A high precision was achieved in the performed tests.