Water Detection through Spatio-Temporal Invariant Descriptors

TL;DR

This paper presents a novel approach for water detection in videos using spatio-temporal descriptors that are invariant to reflections and color variations, validated on a new database and outperforming existing methods.

Contribution

The work introduces a new water detection algorithm based on local descriptors and a new Video Water Database for validation, addressing a less explored area.

Findings

Outperforms existing dynamic texture recognition algorithms by about 5%.

Achieves approximately 15% improvement in material recognition accuracy.

Demonstrates robustness against reflections and color variations in water detection.

Abstract

In this work, we aim to segment and detect water in videos. Water detection is beneficial for appllications such as video search, outdoor surveillance, and systems such as unmanned ground vehicles and unmanned aerial vehicles. The specific problem, however, is less discussed compared to general texture recognition. Here, we analyze several motion properties of water. First, we describe a video pre-processing step, to increase invariance against water reflections and water colours. Second, we investigate the temporal and spatial properties of water and derive corresponding local descriptors. The descriptors are used to locally classify the presence of water and a binary water detection mask is generated through spatio-temporal Markov Random Field regularization of the local classifications. Third, we introduce the Video Water Database, containing several hours of water and non-water videos, to validate our algorithm. Experimental evaluation on the Video Water Database and the DynTex database indicates the effectiveness of the proposed algorithm, outperforming multiple algorithms for dynamic texture recognition and material recognition by ca. 5% and 15% respectively.