[Paper Review] Ubiquitous WLAN/camera positioning using inverse intensity chromaticity space-based feature detection and matching: A preliminary result

Wan Bejuri Wan Mohd. Yaakob, Mohamad Mohd Murtadha+2 more|arXiv (Cornell University)|Feb 27, 2012

Indoor and Outdoor Localization Technologies19 references16 citations

TL;DR

This paper proposes a novel indoor localization method combining WLAN fingerprinting and camera-based feature detection using inverse intensity chromaticity space for robust position estimation. By fusing received signal strength and hallway interest points via model fitting, it eliminates conventional search algorithms, reducing computational complexity and demonstrating promising pre-experimental results in an indoor environment.

ABSTRACT

This paper present our new intensity chromaticity space-based feature detection and matching algorithm. This approach utilizes hybridization of wireless local area network and camera internal sensor which to receive signal strength from a access point and the same time retrieve interest point information from hallways. This information is combined by model fitting approach in order to find the absolute of user target position. No conventional searching algorithm is required, thus it is expected reducing the computational complexity. Finally we present pre-experimental results to illustrate the performance of the localization system for an indoor environment set-up.

Motivation & Objective

• To develop a low-complexity indoor positioning system that avoids traditional search-based localization algorithms.
• To integrate WLAN signal strength measurements with camera-derived interest point features from hallways.
• To improve localization accuracy by fusing multi-sensor data through model fitting.
• To evaluate the feasibility of using inverse intensity chromaticity space for feature detection in indoor environments.

Proposed method

• The system uses inverse intensity chromaticity space to detect and match visual features from hallway scenes captured by a camera.
• Simultaneously, it collects received signal strength (RSS) values from multiple WLAN access points.
• Feature points and RSS fingerprints are combined using a model fitting approach to estimate absolute user position.
• The method avoids iterative searching by directly fitting a position model to the fused data.
• The fusion process leverages spatial consistency between visual features and RSS measurements to refine localization.

Experimental results

Research questions

• RQ1 Can inverse intensity chromaticity space effectively detect and match visual features in indoor hallway environments?
• RQ2 How well can WLAN RSS and camera-based features be fused to improve localization accuracy?
• RQ3 To what extent does the proposed model fitting approach reduce computational complexity compared to conventional search-based methods?
• RQ4 What is the performance of the system in a real-world indoor setup without relying on external positioning infrastructure?

Key findings

• The proposed method successfully reduces computational complexity by eliminating the need for conventional search algorithms.
• The fusion of WLAN RSS and camera-based visual features improves localization robustness in indoor environments.
• Pre-experimental results demonstrate the feasibility of the approach in a real indoor setup.
• The use of inverse intensity chromaticity space enables effective feature detection and matching in hallway scenes.