A hybrid fusion of wireless signals and RGB image for indoor positioning

Abstract

In this article, we propose a new indoor positioning algorithm using smartphones, where wireless signals and images are deeply combined together to improve the positioning performance. Our approach is based on the use of local binary patterns’ feature, which has the advantages of rotation invariance and scale invariance. Moreover, the term “uniform” are fundamental properties of local image textures and their occurrence histogram is proven to be a very powerful texture feature. Besides, the received signal strength acts as a reliable cue on a person’s identity. We first obtain a coarse-grained estimation based on the visualization of wireless signals, which are presented by a vector, making use of fingerprinting methods. Then, we perform a matching process to determine correspondences between two-dimensional pixels and three-dimensional points based on images collected by the smartphone. After being evaluated by experiments, our proposed method demonstrates that the combination of the visual and the wireless data significantly improves the positioning accuracy and robustness. It can be widely applied to smartphones to better analyze human behavior and offer high-accuracy indoor location–based services.