This is a research project completed by Hanchong Jin during Spring 2020 semester under professor Roberto Tron and with the help of PhD student Zachary Serlin at Boston University.

• Random Numbers from 1 to 10
• Two nodes:
  • Node1 handles 1 to 5
  • Node2 handles 6 to 10

• Random Vectors [x1, x2] with -5 <= x1 <= 2, and -5 <= x2 <= 25
• Three Nodes
  • Node0: [0, 0]
  • Node1: [10, 10]
  • Node2: [20, 20]

Only the following two python files are used in this final version.

node.py:

• Extract features from images
  • Run k-means to partition features to each node
  • Keep features of images that are supposed to be handled by itself
  • Publish features of images that are supposed to be handled by other nodes
• Publish collected features to quickmatch_node for processing
  • Every second, the node would publish the features collected to the quickmatch_node and empty its collection

quickmatch_node.py:

• Upon receiving features in the callback function

  • Calculate distance matrix
  • Calculate feature density
  • Build tree
  • Sort edges
  • Break and merge tree
  • Graph results

  Click here for complete documentation.

  Please reach me at hanchong1996@hotmail.com if you are a researcher working on this project and need edit access for the documentation.