Mobile Landmark Recognition - Multimedia Database Retrieval: Technology and Applications - page 129

Chapter 5
Mobile Landmark Recognition

Abstract In recent years, landmark image recognition has been a developing
application for computers. In order to improve the recognition rate for mobile
landmark recognition systems, this chapter presents a re-ranking method. The
query feature vector is modified, identifying important features and non-important
features. These are performed on the ranked feature vectors according to feature
selection criteria using an unsupervised wrapper approach. Positive and negative
weighting schemes are applied for the modification of the query to recognize the
target landmark image. The experimental results show that the re-ranking method
can improve the recognition rate, as compared to previously proposed methods that
utilize saliency weighting and scalable vocabulary tree encoding.

5.1 Introduction

The comparison of photos captured on mobile devices to a landmark photo database
at the remote server is the main issue in mobile landmark search applications. The
mobile user captures a landmark image and then uploads the image data (or a
compact descriptor [ 144 ]) to the server. In an instant, related information of the
captured image is returned to the user, i.e., name, geographic location, photograph
viewpoints, tourism recommendations, or other value added information. This
image matching tool has emerged in applications of mobile phones for not only
landmark recognition, but also mobile shopping, mobile location recognition, online
photographing recommendation, and content-based advertising. This chapter will
focus on mobile landmark recognition, addressing this issue using the re-ranking
method and saliency information, on top of the benchmark of the state-of-the art
method in landmark recognition and retrieval.
In the mobile visual search scenario, instead of sending an entire photo, sending
a compact descriptor computed on the mobile device allows for low bit rate
search. Some compact descriptors are a low dimensional representation of the scale-
invariant feature transform (SIFT) descriptor, such as [ 145 - 148 ]. These research
efforts in compact visual descriptors provide the following benefits. Firstly, the
compact descriptor reduces resource consumption (i.e., less battery and memory
use), since sending large amounts of data via wireless consumes relatively large
mobile resources as compared to sending a compact signature. Secondly, the internet