Daytime Water Detection by Fusing Multiple Cues for Autonomous Off-Road Navigation

Abstract

Detecting water hazards is a significant challenge to unmanned ground vehicle autonomous off-road navigation. This paper focuses on detecting the presence of water during the daytime using color cameras. A multi-cue approach is taken. Evidence of the presence of water is generated from color, texture, and the detection of reflections in stereo range data. A rule base for fusing water cues was developed by evaluating detection results from an extensive archive of data collection imagery containing water. This software has been implemented into a run-time passive perception subsystem and tested thus far under Linux on a Pentium based processor.

Self-supervised segmentation of river scenes

“…The work in [7], [8], [9] uses conventional visible passive imagery to detect bodies of water. In [7] and [8] the authors detect water by fusing multiple cues like color, brightness and stereo depth to train a classifier.…”

“…In [7] and [8] the authors detect water by fusing multiple cues like color, brightness and stereo depth to train a classifier. The recent work in [9] is for monocular vision and looks to generate a physical model of bodies of water lying on or nearby roads.…”

“…Color is a useful feature for detecting water as demonstrated in [7] & [8]. Texture is another useful cue, for example, parts of an image containing the river tend to be less textured than regions of foliage.…”

Context modeling combined with motion analysis for moving ship detection in port surveillance

“…In the literature, it has been shown that the unitary-category classification of Bao 35 demonstrates a better performance compared to another state-of-the-art approach. 46 We have extended the unitary-category classification of Bao et al 35 into multicategory classification and applied contextual information as a special feature. With our contribution, we have shown that our gravity model for region labeling outperforms the results achieved by Bao et al We have evaluated the performance of our gravity model on the WATERVisie dataset.…”

Perception of Environment Properties Relevant for Off-road Navigation

“…The detection of puddles and other water hazards allows the usage of several sensor systems. The camera-based water detection presented in [17] uses color, texture and stereo range data. A polarization-based approach is described in [20].…”