Creating a Streetview for Dioramas UPDATE

I have abandonned the Photo tourism software and I am looking at “Image connectivity graph”

Once all pairs of images have been matched, we can construct an image connectivity graph to represent the connections between the images in the collection. An image connectivity graph contains a node for each image, and an edge between any pair of images that have matching features.

To create this visualization, the graph was embedded in the plane using the neato tool in the Graphviz graph visualization toolkit. Neato works by modeling the graph as a mass-spring system and solving for an embedding whose energy is a local minimum.

The image connectivity graph for this collection has several notable properties. There is a large, dense cluster in the center of the graph that consists of photos that are mostly wide-angle, frontal, well-lit shots of the fountain [such as (a)]. Other images, including the “leaf” nodes [such as (b) and (c)] corresponding to tightly cropped details, and nighttime images [such as (d)], are more loosely connected to this core set.

Image connectivity graph for the Trevi Fountain. This graph contains a node (red dot) for each image in a set of photos of the Trevi Fountain, and an edge between each pair of photos with matching points. The size of a node is proportional to its degree. There are two dominant clusters corresponding to daytime photos e.g., image (a) and nighttime photos image (d). Similar views of the facade are clustered together in the center of the graph, while nodes in the periphery of the graph, e.g., (b) and (c), are more unusual (often closeup) views.

Image connectivity graph for the Trevi Fountain. This graph contains a node (red dot) for each image in a set of photos of the Trevi Fountain, and an edge between each pair of photos with matching points. The size of a node is proportional to its degree. There are two dominant clusters corresponding to daytime photos e.g., image (a) and nighttime photos image (d). Similar views of the facade are clustered together in the center of the graph, while nodes in the periphery of the graph, e.g., (b) and (c), are more unusual (often closeup) views.

there are few open question to answers:

I am going to do testing with a image dataset from Cyan’s Myst and Riven games1

  1. I own the games from gog.com, I have DVD/CD-ROMs somewhere2 

  2. Do I, the may have gone out in the clean out after the Queensland 2010-11 flood 

> SELECT * FROM "MetaData" WHERE post_slug = "/notebook/2018/11/streetview-Diorama-update"

Author: Tom Sparks
DateTime: 

syndication:

OK >
--- pagination: enabled: true ---