Home // International Journal On Advances in Software, volume 12, numbers 3 and 4, 2019 // View article

Discovering Hotspots with Photographic Location and Altitude from Geo-tagged Photograph

Authors:
Masaharu Hirota
Jhih-Yu Lin
Masaki Endo
Hiroshi Ishikawa

Keywords: area of interest; density-based clustering; grid-based clustering; photograph location; clustering.

Abstract:
A hotspot is an interesting place where many people go sightseeing. A place where many photographs have been taken (which we call a hotspot) might be an interesting place for many people to visit. Analyzing such places is important to promote industries such as those related to tourism. To identify hotspots, most existing research applies a grid-based or density-based clustering algorithm, such as density-based spatial clustering of applications with noise (DBSCAN) or mean shift. When applying such methods to hotspot detection, the features used for clustering are latitude and longitude. Therefore, the identified hotspots are visualized in a two-dimensional space. However, large areas, landmarks, and buildings may include elevated hotspots or multiple hotspots with different altitudes, which cannot be distinguished by latitude and longitude. Therefore, in this research, we propose methods for identifying hotspots based on altitude, in addition to latitude and longitude, and visualizing these hotspots in a three-dimensional space. We propose two types of method, based on density-based and grid-based clustering, that use these features. The first method is one that improves ST-DBSCAN, which clusters data based on spatial and time features. The other method is an extension of general grid-based clustering using these features. As an example application, we classified the identified hotspots as shooting spots, observation spots, areas of interest, and others. We demonstrate our approach by identifying hotspots in a three-dimensional space using photographs obtained from Flickr, and discuss the usefulness of detecting hotspots using altitude in addition to latitude and longitude.

Pages: 322 to 331

Copyright: Copyright (c) to authors, 2019. Used with permission.

Publication date: December 30, 2019

Published in: journal

ISSN: 1942-2628