02.2 Image collection algorithm

Node

A node is a spatial indexed point. Some urban features like hospitals, pharmacies, bus stops, etc. can be described as a node. The steps taken to collect an image from a node is:

1. Get the node(N) coordinates (i.e. latitude (LAT) and longitude(LNG))
2. Search inside a radius(R), centered at the node coordinates, for available panoramas (i.e. Google Street View Panoramas).
3. Select arbitrarily for an panorama(P).
4. Get an image from the panorama P heading towards the node N, that is, with the heading property pointing from the panorama coordinates towards the node N coordinates.

Segments

The steps below assume that panoramas are collected from the Google Street View API

Segments are spatially coherent sequences of nodes (i.e. a street is composed of nodes aligned in some straight or curved line). The steps to take images from a sequence are:

1. For each node(N) in the sequence:
   1. Get a panorama for node N inside a radius(R). Note that the size of R is related to the sampling rate of the segment, that is, a large R may result in the same panorama being found for different nodes and a small R may result in no panoramas being found at all.
   2. For the first retrieved panorama P for the node N get an image heading toward the vehicle front direction (collected from the panorama).