GeoSIM News:

 

GeoSIM: An Urban Sensing System for Social Image Mapping of Urban Geolocations

With current advances in mobile technology, generating the texture information from the images taken by users from an urban environment is becoming increasingly feasible. We envision GeoSIM (Geo Social Image Mapping) as a system with which a group of individuals with camera equipped mobile phones participate in collaborative/social mapping of such texture information in a virtualized urban area.

 

Suppose the 3D model of the urban environment at a target geolocation is available (e.g., through Google EarthTM), but the environment texture is missing. With GeoSIM, one can define a distributed urban image documentation project to capture the texture of the environment through social participation as follows. GeoSIM server enrolls the participants as they join the project by running the GeoSIM client on their mobile phones. Once enrolled, the participant identifies her preferences/constraints (e.g., desired starting point and moving direction, available time, etc.) through the client. Subsequently, the GeoSIM server directs her by providing an individualized participation plan consisting of: 1) an optimal navigation path through the urban area with designated stops for image acquisition, and 2) a set of instructions for image acquisition at each stop (e.g., target object, camera settings and orientation, desired lighting). The navigation path should be optimized across all participants for efficient, fast and comprehensive documentation of the urban image, while satisfying the constraints and preferences of the participants. Meanwhile, the acquired images (i.e., the participatory sensed data) are collected in an ongoing basis by the GeoSIM server, processed, and merged for progressive texturing of the 3D model of the urban environment.

 

To consider the dynamics of the urban image in multiple resolutions, GeoSIM considers the spatial resolution and the timestamp of the acquired images to simultaneously create/document several distinct instances of the urban image with different spatial resolutions at different times. In addition, each instance of the urban image comes in multiple versions, where each version is composed of the images with relevant tags (e.g., New Year’s Eve, During Civil Unrest), such that users can browse the created urban images according to their desired tags.

Proposed Architecture

The figure below depicts the architecture we propose for GeoSIM as an application running on top of  GeoDec, a scalable geospatial data management engine developed at  Infolab.

 

 

 

GeoSIM consists of server and client side modules as follows:

 

GeoSIM Server Modules

GeoSIM Client Modules
GeoDec+ Although the capabilities of the GeoDec components are useful, they are not sufficient to implement GeoSIM. Therefore, to implement GeoSIM we first need to extend GeoDec to, what we call, GeoDec+. GeoDec+ Lite

 

 It is a reduced version of GeoDec+ and enables users to access and query the virtualized geolocation from their mobile phone.
Coordination Engine Plans the GeoSIM users participation for efficient, fast and comprehensive social image mapping, while ensuring that the users constraints and preferences are satisfied.
Data Collection Server Along with the data collection client, data collection server allows collection of the acquired data/image from the user mobile phones into a central repository. Data Collection Client Paired with the data collection server, it enables collection of the participatory data
Texture Generation It uses the images generated by users to evolve the texture of the urban environment.

 

 

Challenges

  • Scale and dynamism of the user group
  • Dynamic texture over time
  • Path planning and visibility analysis integration
  • Using uncaliberated images for texture generation
  • Texture generation depending on the available preexisting data

 

 

 

Sponsors :

 

 

 

 

 

  
webadmin: beipan@usc.edu