Archival and Retrieval of Lost Objects using Multi-feature Image Matching in Mobile Applications

Abstract

The rapid development of mobile computing technologies, as well as enhanced wireless communications, has paved the way for the development of the e-Government and m-Government systems. With the aim of increasing the accessibility of government services, such systems are still proven to be under-utilized, especially in the area of retrieving lost personal objects. In this paper, we propose a multiplatform mobile application for reporting and retrieval of lost objects in an efficient manner, rather than going through a manual procedure of filling up forms. The backend sever runs an object retrieval algorithm on the reported lost and found objects in the database, which is comprised of a textual search, a geographic information system filter, an image matching algorithm using Speeded Up Robust Features, followed by color validation using L*a*b* color space. Experiments have been performed on 40 lost objects against the database of 50 found objects with the accuracy of 95%, while providing a precision value of 100% and a 90% of recall. The efficiency and overall good performance of the proposed system can help in reducing the manual labor, as well as in providing fast feedback to the users.