Article ID Journal Published Year Pages File Type
4947064 Neurocomputing 2017 14 Pages PDF
Abstract
In recent years, many approaches have been proposed to compensate the lack of performance of GNSS (Global Navigation Satellites Systems) occurring when operating in constrained environments. One of these approaches consists in characterizing the environment of reception of GNSS signals using a wide-angle (fisheye) camera oriented to the sky. The content of acquired images is classified into two regions (sky and not-sky) in order to determine LOS (Line-Of-Sight) satellites and NLOS (Nonline-Of-Sight) satellites. This paper is aimed at proposing an image-content classification method to make this approach more effective. The proposed method is composed of four major steps. The first one consists of simplifying the acquired image with an appropriate couple of colorimetric invariant and exponential transform. In the second step, the simplified image is segmented using Statistical Region Merging method. The third step consists of characterizing the segmented regions with a number of local image region descriptors providing more statistically meaningful and discriminatory features. In order to classify the characterized regions into sky and non sky regions, we propose the supervised MSRC (Maximal Similarity Based Region Classification) method by using Bhattacharyya coefficient-based distance. Comparative and extensive experiments have been conducted to investigate the effectiveness of the proposed MSRC method according to the proposed groups of local image region descriptors. Furthermore, we clearly validate the feasibility of MSRC method by comparing its results with those presented in the state of the art.
Related Topics
Physical Sciences and Engineering Computer Science Artificial Intelligence
Authors
, , , , , , , ,