| Package Name: Segmentation_2D_plus |
|
| Optimal segmentation algorithm for data in spaces of dimension 2 or higher. |
| Brad Jackson, Jeff Scargle, and Jay Norris |
|
| Abstract:
The problem of detecting statistically significant structure in image data can be solved by obtaining the partition that maximizes the fitness of a nonparametric model of the data. This paper presents a simple procedure for carrying out this procedure by transforming the 2D (or higher) problem to 1D, determining the optimal block structure using our previously developed 1D algorithm, and then returning to the 2D representation to reconstruct the image structures.
|
|