Visual Localizer: Outdoor Localization Based on ConvNet Descriptor and Global Optimization for Visually Impaired Pedestrians.

Shufei Lin,Kailun Yang,Kaiwei Wang,Ruiqi Cheng

doi:10.3390/s18082476

Abstract

Localization systems play an important role in assisted navigation. Precise localization renders visually impaired people aware of ambient environments and prevents them from coming across potential hazards. The majority of visual localization algorithms, which are applied to autonomous vehicles, are not adaptable completely to the scenarios of assisted navigation. Those vehicle-based approaches are vulnerable to viewpoint, appearance and route changes (between database and query images) caused by wearable cameras of assistive devices. Facing these practical challenges, we propose Visual Localizer, which is composed of ConvNet descriptor and global optimization, to achieve robust visual localization for assisted navigation. The performance of five prevailing ConvNets are comprehensively compared, and GoogLeNet is found to feature the best performance on environmental invariance. By concatenating two compressed convolutional layers of GoogLeNet, we use only thousands of bytes to represent image efficiently. To further improve the robustness of image matching, we utilize the network flow model as a global optimization of image matching. The extensive experiments using images captured by visually impaired volunteers illustrate that the system performs well in the context of assisted navigation.

Highlights

In the world, 253 million people are estimated to be visually impaired, of whom 36 million people are totally blind [1]
Aiming to address the problems of viewpoint, appearance and route changes on visual localization, we propose a novel visual localization system—Visual Localizer
To achieve robust image representation, different layers derived from five prevailing ConvNets are evaluated on their robustness against various environmental changes

Summary

Introduction

253 million people are estimated to be visually impaired, of whom 36 million people are totally blind [1]. The majority of visually impaired people, especially those in China, still use simple and conventional assistive tools, e.g., white canes. The extreme lack of assisted navigation approaches is not a rare situation among visually impaired people. According to our long-term observation and investigation, one of the most urgent demands for people with impaired vision lies in outdoor navigation with the goal to reach their destinations. Thanks to the proliferation of intelligent devices and mobile Internet, the visually impaired people get access to coarse GNSS localization using mobile navigational applications on any ordinary smart phone. The errors of GNSS positioning is not so critical for sighted people, in that the visual capability helps to localize themselves and to reach the desirable place. Things go differently for the visually impaired people. Imagining a common scenario that a person with visual impairments stands at the vicinity of a turning, it is tough for Sensors 2018, 18, 2476; doi:10.3390/s18082476 www.mdpi.com/journal/sensors

Objectives

Results

Conclusion