Temporally stable feature clusters for maritime object tracking in visible and thermal imagery

Abstract

This paper describes a new approach to detect and track maritime objects in real time. The approach particularly addresses the highly dynamic maritime environment, panning cameras, target scale changes, and operates on both visible and thermal imagery. Object detection is based on agglomerative clustering of temporally stable features. Object extents are first determined based on persistence of detected features and their relative separation and motion attributes. An explicit cluster merging and splitting process handles object creation and separation. Stable object clusters are tracked frame-to-frame. The effectiveness of the approach is demonstrated on four challenging real-world public datasets.

Knowledge-Driven Semantic Segmentation for Waterway Scene Perception

“…Ćesić et al [28] compared images of obstacles detected by visual detection using a stereo camera with those detected by a radar detection technique. Osborne et al [29] used visible and thermal imaging with time-domain stability characteristics to separate and classify targets. Wang et al [30] integrated stereo photography with visual obstacle separation technology to detect obstacles below the coastline.…”

Stereo obstacle detection for unmanned surface vehicles by IMU-assisted semantic segmentation

“…In practice, however, this boundary is often blurred due to unfavorable weather conditions (haze, overwhelming cloudiness, fog), sun glitter, and reflections of the surrounding environment in water, making estimation of the exact position of the horizon difficult. Osborne et al [23] propose a method capable of tracking objects on both visible and thermal imagery. They detect obstacles based on agglomerative clustering of temporally stable features and track stable object clusters frame-to-frame.…”