Using Particle Measure
From Endrov and VWB
One of the most important data extraction functions is Particle Measure. The use of it is exemplified below:
Contents |
Choosing objects
To use it, you need:
- A channel with the regions you are interested in (input: regions). Each region has a unique color.
- Optionally an image that you want to quantify (input: values). If you only want geometrical information, use regions as values
To find the regions you first need to segment the image. It is not part of this tutorial to teach how to do that and this image assumes it has been done (1). Typically your input will be a boolean (0 and 1) stack, depicting regions. To separate these regions (give them unique values), use identify particles (2). It is possible to ignore the 0-valued object which typically is the segmented out uninteresting region.
Extraction
Use (3) to extract parameters. These are collected on a stack-per-stack basis. If you want to associate regions over time you should an alternative "identify particles" or do it as a post-processing step. For the sake of speed you have to select the interesting parameters.
Showing
The output can be shown (4) inside the flow. For convenience of execution it can help to use the output port.
Storage
The data can be stored either in a CSV file (6) or in an SQL database (5). CSV is easiest for beginners while SQL is more convenient for large amounts of data.
CSV
CSV can be imported in most spreadsheet programs e.g. OpenOffice. The character between columns can be changed but tabs are usually good.
SQL
To use SQL you first have to set up a server. PostgreSQL is recommended, MySQL less so. Any JDBC-compliant SQL database can be used. You may have to drop new .jar-files into the libs/-directory, see the instructions for your database (let us know if you have problems). You have to create a new database yourself, and from that you will know the URL, user and password. The dataid parameter is used to store the output of multiple recordings in the same table; there will be a separate dataid column that can be used to distinguish them.
There are buttons to manually interact with the database. You need not bother with these normally (you have to execute before they work) unless you change which parameters are extracted.
Filtering
TODO
Interaction
TODO
