Skip to content

Latest commit

 

History

History
26 lines (17 loc) · 1.54 KB

README.md

File metadata and controls

26 lines (17 loc) · 1.54 KB

cluster-detection

[![View my LinkedIn Profile at https://es.linkedin.com/in/dgfguida] (https://static.licdn.com/scds/common/u/img/webpromo/btn_liprofile_blue_80x15.png)] (https://es.linkedin.com/pub/diego-guida-f%C3%B3rneas/109/531/781)

Octave algorithms used for automatically detecting clusters in stacks of low-contrast images and tracking growth events such as Nucleation, Expansion and Coalescence.

Binarization.m

Algorithm used to detect clusters in a gray image. Clusters of information should have a higher level of intensity than background otherwise specify if the opposite is true by setting the parameter "inverted" to true.

I/O:

  • Input: Folder containing .tif images to analized.
  • Output: Each image under a new folder "Final".

DomainAnalisis.m

Algorithm used to follow cluster evolution through the images. Nucleation, Expansion and Coalescence events are tracked by comparing the images. Images should be alphabetically ordered by default.

Events:

  • Nucleation: When a cluster appears where there was none before. Has to remain for NuclThresh Nº of images to be considered. Otherwise discarded as noise.
  • Expansion: When a cluster total area increases above the Error_Area parameter. It is considered to have increased size and therefore expanded.
  • Coalescence: When two or more clusters suddenly merge together into one. From that point on, the clusters are to be considered a single one.

I/O:

  • Input: Folder containing the alphabetically ordered images.
  • Output: Table with the Nº of events per image as well as the total cluster area.