Until a few years ago, the spatial information produced by Rearview Mirror sensors was managed and analysed using dedicated software (GIS) in file-based data formats (e.g. shapefiles). Nowadays, the most advanced approaches in data management consider the spatial component of objects (e.g. a moving car) as one of its many attributes: thus, while understanding the spatial nature of your data is essential for proper analysis, from a software perspective, spatial is (increasingly) not special. Spatial databases are the technical tools needed to implement this perspective. They integrate spatial data types (vector and raster) together with standard data types that store the objects’ other (non-spatial) associated attributes. Spatial data types can be manipulated by SQL through additional commands and functions for the spatial domain. This possibility essentially allows you to build a GIS using the existing capabilities of relational databases. Moreover, while dedicated GIS software is usually focused on analyses and data visualisation, providing a rich set of spatial operations, few are optimised for managing large spatial data sets (in particular, vector data) and complex data structures. Spatial databases, in turn, allow both advanced management and spatial operations that can be efficiently undertaken on a large set of elements. This combination of features is becoming essential, as with car movement data sets the challenge is now on the extraction of synthetic information from very large data sets rather than on the extrapolation of new information from limited data sets with complex algorithms.
you will extend your database with the spatial dimension of Rear View Mirror Camera locations and start to familiarise yourself with spatial SQL. You will implement a system that automatically transforms coordinates from a pair of numbers into spatial objects. You are also encouraged to explore the PostGIS documentation where the long list of available tools is described. At the moment, your data are stored in the database and the GPS positions are linked to individuals. While time is correctly managed, coordinates are still just two decimal numbers (longitude and latitude) and not spatial objects. It is therefore not possible to find the distance between two points, or the length of a trajectory, or the speed and angle of the step between two locations. In this chapter, you will learn how to add a spatial extension to your database and transform the coordinates into a spatial element (i.e. a point).
It is recommended that you not store Rearview Mirror Camera position data in some projected reference system, but instead keep them as longitude/latitude WGS84. You can later project your features in any other reference system whenever needed. There are two options: they can be stored as geography data type or as geometry data type, specifying the geographic reference system by its SRID code, which in this case is 4236. The natural choice for geographic coordinates would be the geography data type because the geometry data type assumes that geographic coordinates refer to Euclidean space. In fact, if you calculate the distance between two points stored as geometry data type with SRID 4326, the result will be wrong (latitude and longitude are not planar coordinates so the Euclidean distance between two points makes little sense)
More information at http://www.jimilab.com/