cyrildiagne / RasterFairy

The purpose of Raster Fairy is to transform any kind of 2D point cloud into a regular raster whilst trying to preserve the neighborhood relations that were present in the original cloud. A typical use case is if you have a similarity clustering of images and want to show the images in a regular table structure.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

RasterFairy

The purpose of Raster Fairy is to transform any kind of 2D point cloud into a regular raster whilst trying to preserve the neighborhood relations that were present in the original cloud. A typical use case is if you have a similarity clustering of images and want to show the images in a regular table structure.

Requirements

  • numpy > =1.7.1
  • scipy - only for coonswarp and rfoptimizer

Installation

Version 1.02 is not at proper Python package yet, so just copy the files

rasterfairy.py
prime.py
coonswarp.py 
rfoptimizer.py 

into your application folder to use it

Usage

Basic usage:

import rasterfairy

#xy should be a numpy array with a shape (number of points,2) 
grid_xy = rasterfairy.transformPointCloud2D(xy)
#grid_xy will contain the points in the same order but aligned to a grid

To-Do

  • Add hexagonal circle grid
  • Look into further improving splitting process

Related Projects

As I learned after publishing there is a very similar earlier developed technique called IsoMatch by O. Fried, S. DiVerdi, M. Halber, E. Sizikova and A. Finkelstein. Unfortunately I was not aware of it during my research and their solution works differently, but you might want to check it out and see if it's better suited to your requirements: IsoMatch Codebase

Another related technique is Kernelized Sorting by Novi Quadrianto, Le Song, Alex J. Smola. from 2009 Kernelized Sorting

Kyle McDonald's CloudToGrid project is a Python-based implementation of the Hungarian method.

A note about porting this to other languages

If you want to port this algorithm to another language like C++, Javascript or COBOL I'm very happy about it. Only there is a little thing about "porting etiquette" I want to mention - yes, it will take you some work to translate those 500+ lines of code into the language of your choice and you might have to change a few things to make it work. Nevertheless, the algorithm stays the same and yes - I'm probably quite vain here - but I like to read my name. In big letters. Bigger than yours. And don't even think about writing anything like "insipired by". So the proper titling for a port will read something like "Raster Fairy by Mario Klingemann, C++ port by YOU".

ezoic increase your site revenue

About

The purpose of Raster Fairy is to transform any kind of 2D point cloud into a regular raster whilst trying to preserve the neighborhood relations that were present in the original cloud. A typical use case is if you have a similarity clustering of images and want to show the images in a regular table structure.

License:BSD 3-Clause "New" or "Revised" License


Languages

Language:Python 100.0%