Page history Edit this page How do I edit this website?
Original MediaWiki page
This page describes content relating to the ImageJ2 platform. Click the logo for details.

Jython Scripting

The content of this page has not been vetted since shifting away from MediaWiki. If you’d like to help, check out the how to help guide!

This page describes how to write scripts in Jython, the JVM-based flavor of Python. To call ImageJ functions from Python programs, see PyImageJ.

Introduction

Jython is an implementation of the Python programming language designed to run on the Java platform. 1 In ImageJ, Jython is one of several supported languages.

When to use Jython

All scripting language supported by ImageJ can be used to access the ImageJ API. There are only differences in how the imports are handled and in the syntax of the selected language. Jython has a syntax that differs from most other language as indentations instead of brackets are used to group code blocks.

The following list will help you to decide if Jython is the right choice to create scripts for ImageJ:

  • If you have experience with Python, you can easily use Jython for ImageJ scripting. But you have to keep in mind that tools commonly used in many Python projects (e.g. Numpy) are not available in Jython. By building your own modules you can create complex scripts that otherwise are only possible by writing ImageJ plugins in Java.
  • If don’t have any experience in programming, the Python language is a good choice to start with. If your only aim is to write scripts for ImageJ, there are other languages you should try first (e.g. Groovy).
  • In Python, many problems can be solved with less code than in other languages. Nonetheless, the code is easy to read. Have a look at the examples on this page and decide if you want to start using Python for ImageJ scripting.

Explanation

The Java implementation of Python is limited to the standard library of Python 2.
It is not possible to use external python modules (like Numpy…) however, any Java class residing in the Fiji installation can be used.
Even with the given limitations, Jython is a powerful language for ImageJ scripting. Hopefully the examples on this page can convince you of that.

Jython basics for ImageJ

For an introduction in ImageJ scripting visit the page Scripting basics.

Introduction

The aim of this page is not to teach how to program in Python. This purpose is much better fulfilled by the documentation of Python 2. The focus of this page is to show how features of the Python language can be useful for ImageJ scripting.

That is why more complex examples are used that are fully functional. Just copy the code to the Script Editor and try them by yourself. Extensive in-line documentation is used to explain the implementation.

Hello World

- With print

There are 2 ways to print some information back to the user.
The first one is a classical python print statement, that will print some information to the console.
print "Hello world"
You can print any kind of variable and objects.

print "This is a string followed by an int", 10

NB1 : If used in a plugin, and no console window is open then the printed information will not be visible to the user (contrary to the log function below)

NB2 : Using numerous print statements might slow down the execution time when used in a plugin (not observed when executing from the script interpreter).

- With IJ.log()

from ij import IJ
IJ.log("Hello world")
IJ.log("This is a string followed by an int " + str(10))

Contrary to the print statement the log function display some output into a log window (newly open if not already open), and accept only a string as argument.

Image selection using the GenericDialog class

This example script will create up to 10 new images and create a GenericDialog to select 3 of them. Finally the names of the selected images are printed to the Log window. It is recommended to copy the code to the Script Editor and run it by yourself.

The following list links to documentation of the used Python features:

Using Scripting Parameters

The second example is inspired by atomic resolution images recorded with an Transmission Electron Microscope (TEM). Such images show a regular structure (a crystal), but the images are noisy because of the low signal. By using a Fourier filter the contrast can be enhanced.

The script will create a periodic structure and add some random noise. The user can control the parameters of the created image. This is realized using Script parameters. The Fourier filtering has been created by using the Recorder. Finally a simple image calculator is used to show that functions can be passed as parameters.

This list links to the documentation of Python features that are introduced with this example:

A batch opener using os.walk()

We have yet introduced some powerful functions build into Python. Another one is walk() from the os module. It can be used to go through a directory structure and process the contained files. In this example walk() is used to batch open images with ImageJ’s function openImage().

To read more about the used features, the following list provides links to additional information:

Importing Java module and classes

Another great feature of Jython is the possibility to use functions from Java jar package that resides in the jar folder of imageJ.

ImageJ and Fiji API

The following API documentation lists all available modules and functions :

Those package are built-in with Fiji, but any package that resides in the jars folder can be imported provided you know the path to the class.

For example, one of the main built-in ImageJ packages is called ij, and often Jython scripts will write something like this at the top:

from ij import IJ
# do stuff below....

Doing this allows you to access the IJ class which resides in the ij package. You can find a description of the ij package here. What can we do with the IJ class? Clicking on the IJ link brings you to the class documentation page for IJ. This class contains “static utility methods” which means you can call them with without instantiating (calling the constructor) the IJ class. We will cover constructors later. Looking through the documentation for IJ, lets focus on the method createImage (docs here). This method can be called just like you would call a method on a python class. The documentation shows you need to provide the following parameters (types in parenthesis):

  1. title (string)
  2. width (int)
  3. height (int)
  4. depth (int)
  5. bitdepth (int)

and it returns an ImagePlus object. ImagePlus objects are very important in ImageJ, and you will the documentation for them here. Below is an example of how to import and use the static methods on the IJ class to create an image.

from ij import IJ # read this as: "from the ij package import the IJ class"
test_img = IJ.createImage("Test image", 512, 512, 1, 8)
# now check the type of test_img
print(type(test_img))
# <type 'ij.ImagePlus'>

This code shows that we have successfully created an ImagePlus object. Looking at the documentation for the ImagePlus class, let’s use a few of the methods to make sure the image was created correctly.

from ij import IJ
test_img = IJ.createImage("Test image", 512, 512, 1, 8)
# check the type:
print(type(test_img))
# <type 'ij.ImagePlus'>
title = test_img.getTitle()
width = test_img.width
height = test_img.height
print("{} is {} wide and {} tall.".format(title, width, height))
test_img.show()

We accessed the title using the getTitle() method, which takes no arguments and returns the image name. We accessed the image width and height by accessing test_img’s fields. These are not methods, but contain information about the class. We could have also used the getWidth() and getHeight() methods as well. We then called the show() method on our test image and a (very boring) 512X512 8 bit image should have popped up.

Here is another example where we use the ImageJ package and the RoiManager class. According to the javadoc, the RoiManager class resides in ij.plugin.frame. Therefore the code will look like :

from ij.plugin.frame import RoiManager
RM = RoiManager()        # we create an instance of the RoiManager class
rm = RM.getRoiManager()  # "activate" the RoiManager otherwise it can behave strangely

Using openCV in Jython

It is even possible to use most of opencv functionalities within Jython/Fiji. There are several options (see the wiki page about opencv), yet the most straight forward is probably IJ-OpenCV which is available via the update sites. It will automatically download the necessary packages and dependencies in your Fiji installation.

A manual installation is also possible by putting the jar packages in the jar folder of imageJ. They are available on the IJopenCV github, which even provides a maven option.

Matrices

The first thing to know about OpenCV is that most functions work with an OpenCV matrix object. Fortunately, the IJ-OpenCV project provides some converters :

#@ ImagePlus ImP
from ijopencv.ij      import ImagePlusMatConverter
from ijopencv.opencv  import MatImagePlusConverter
from ij               import ImagePlus

# Convert ImagePlus (actually the contained ImageProcessor) to Matrix object
imp2mat = ImagePlusMatConverter()
ImMat = imp2mat.toMat(imp.getProcessor())
print ImMat

# Convert Matrix object to ImageProcessor
mat2ip = MatImagePlusConverter()
NewIP  = mat2ip.toImageProcessor(ImMat)
NewImp = ImagePlus("Matrix converted back to ImagePlus", NewIP)
print NewImP

Such kind of converter is also available for PointRoi to opencv keypoints…

Now to use opencv function, we use the JavaCPP API that contains almost all functions of opencv.

from org.bytedeco.javacpp.opencv_core   import Mat, CvMat, vconcat

## Typical matrices ##

# Identity Matrix of size (3x3) and type 8-bit
Id = Mat().eye(3,3,0).asMat()
print Id
print CvMat(Id) # handy to visualise the matrix

# Matrix of ones (3x3)
One = Mat().ones(3,3,0).asMat()

# Matrix of zeros (3x3)
Zero = Mat().zeros(3,3,0).asMat()

# Custom Matrices
# 1D-Matrix can be initialize from a list
# For 2D (or more) we have to concatenate 1D-Matrices

Row1 = Mat([1,2,3,4,5]) # 1D matrix
Row2 = Mat([6,7,8,9,10])

TwoColumn = Mat()              # initialise output
vconcat(Col1, Col2, TwoColumn) # output stored in TwoColumn
print CvMat(TwoColumn)

<div class="notice" style="font-size: 2; background: #ffcccb; border-left: 10px solid #f57900"><div class="notice-icon"><img src="/media/icons/warning.png"></div><div class="notice-content"><p>The <code class="language-plaintext highlighter-rouge">org.bytedeco.javacpp.opencv_core.Mat</code> object is different than the <code class="language-plaintext highlighter-rouge">org.opencv.core.Mat</code> !! They don’t have exactly the same attributes and functions. In Fiji you should always use the objects from <code class="language-plaintext highlighter-rouge">org.bytedeco.javacpp</code>.</p>
</div>
</div>

Similarly there is some apparent redudancy for the function in the javacpp API.

ex : Transform exists in 3 different places :

  • org.opencv.core.Core.transform

This one takes org.opencv.core.Mat as input. It is currently challenging to have such object in Fiji.

  • org.bytedeco.javacpp.opencv_core.cvTransform

using CvArr as input, but even if you manage to convert your input as a CvArr it crashes Fiji. Apparently it is a deprecated version.

  • org.bytedeco.javacpp.opencv_core.transform

That’s the one to use ! It takes only org.bytedeco.javacpp.opencv_core.Mat as input, which is the most appropriate in Fiji/Jython

Scalar

In addition to Matrices, opencv allows to use Scalar objects A scalar is a 4 item element (v0, v1, v2, v3). If v1=v2=v3=0 then the Scalar is real.

from org.bytedeco.javacpp.opencv_core   import Scalar

# Real scalar can be initiated with a float parameters
Number = Scalar(5.0)
Number = Scalar(float(5))
print Number

# Using an integer as parameter has a different meaning
Empty = Scalar(5) # This initiate an empty Scalar object of size 5
print Empty

# Alternatively one can set the other values of the Scalar
Complex = Scalar(1,2,3,4)
print Complex

Operations

It is possible to perform some operations between matrices, or between Scalar and matrices.

from org.bytedeco.javacpp.opencv_core   import Scalar, Mat, subtract

A = Mat([1,2,3,4,5])
B = Mat([1,2,-3,-4,0])

Number = Scalar(10.0)

## Number - B ( B-Number is also possible)
Expr = subtract(Number,B)
print CvMat(Expr.asMat())

## A - B
Out = Mat()
subtract(A,B,Out)
print CvMat(Out)

Self written Jython modules for ImageJ

In Jython you can write all commands line by line in a single file and execute it. To create a neat program, functions and classes can be used to structure code. To prevent using copy&past for regularly used functions and classes, modules are the way to choose. Modules are files that contain functions and classes to import into other files.

To load modules, one has to save them to a directory where Jython will find them. Two lines of code will reveal these directories to you:

from sys import path
print(path)

When running this code the result is an output like

['/home/michael/Software/ImageJ.app/jars/Lib', '/home/michael/Software/ImageJ.app/jars/jython-shaded-2.7.0.jar/Lib', '__classpath__', '__pyclasspath__/']

This tells us that the folder jars/Lib/ inside our ImageJ/Fiji directory is the right place to save modules. As Lib/ does not exist by default, we have to create it.

When a module is imported for the first time, Jython will compile it to Java code. If there is a module named myModule.py, Jython will create a file called myModule$py.class. The next time the module is imported, the jython interpreter uses the .class file instead of the .py file, even if this .py file was modified.

To force the interpreter to use the last version of the py script there are 2 possibilities :

  • Close Fiji, delete the myModule$py.class file and restart Fiji
  • Use the following lines of code (found at stackoverflow) that will force Jython to recompile all modules
# Use this to recompile Jython modules to class files.
from sys import modules
modules.clear()
# Imports of Jython modules are placed below:
import myModule

Adding a custom directory

If you don’t want to use jars/Lib/ to save your modules, you have to extend the array sys.path:

from sys import path
from java.lang.System import getProperty

# extend the search path by $FIJI_ROOT/bin/
# 'fiji.dir' works for plain ImageJ, too.
path.append(getProperty('fiji.dir') + '/bin')
# an alternative can be the users home directory
# path.append(getProperty('user.home') + '/JythonModules')

# Now you can import $FIJI_ROOT/bin/myModule.py
import myModule

The function getProperty() accepts many more strings. A list can be found at The Java Tutorials.

Self written Jython packages for ImageJ

On the way to perfectly organize Jython code, packages are the next step. A Jython package is a folder that contain a set of modules scripts together with a __init__.py file. This file can be empty. Below is a typical structure for the ImageJ.app/jars/Lib folder:

    ImageJ.app/jars/Lib/
    -- myModule.py
    -- myPackage/
       -- __init__.py
       -- mathTools.py
       -- customFilters.py
       -- fftTools.py
    -- myPackage2/
       -- __init__.py
       -- mathTools.py
       -- stackProcessing.py

There are two packages and one module. The first package contains three modules and the second package contains two modules. We can import the modules on different ways:

# Import the single module using the default name:
import myModule

# Import mathTools from the first package
import myPackage.mathTools
# Use a function from the imported module
myPackage.mathTools.aFunction()

# Import mathTools from the second package
from myPackage2 import mathTools
# Use a function from the imported module without prefixing the package
mathTools.aFunction()

# Import customFilters from the first package and rename it
from myPackage import customFilters as filters
# Use a function from customFilters.py
filters.aFunction()

# Importing all module from a package
from myPackage2 import *
# The next line will fail
stackProcessing.aFunction()

The reason for the last import to fail is the empty __init__.py. We have to define which modules of the package are imported when using import *. This is done by setting the variable __all__ in the __init__.py. For myPackage2 this line of code is needed:

__all__ = ["mathTools", "stackProcessing"]

Besides setting this variable, the __init__.py file can contain normal Jython code that is executed upon import of the package.

Bundle packages in a JAR file

An interesting feature of Jython is to search for packages and modules inside of JAR (file format). The folder structure from the last section can be modified by packing everything into a single myPackages.jar. The name of the JAR file doesn’t matter. All imports work the same as explained before.

    ImageJ.app/jars/Lib/
    -- myPackages.jar
       -- myModule.py
       -- myPackage/
          -- __init__.py
          -- mathTools.py
          -- customFilters.py
          -- fftTools.py
       -- myPackage2/
          -- __init__.py
          -- mathTools.py
          -- stackProcessing.py

The advantage of this approach is that you can share your packages easily. For example you can upload the JAR file to an update site. It is possible to upload .py scripts to update sites too, without packaging into a jar. The advantage of jar are that they allow to define dependencies more systematically.

Contrary to the scripts in Jars/Lib, the menu macro scripts are not compiled, and as explained above they can not be imported in other scripts since the Plugin folder do not reside in the Jython search path by default.

This is the reason why a given project is rather distributed in 2 different jar files as explained here.

Using maven to build packages

Using maven you can automate the packaging of Jython code into JAR files. This approach is only recommended if you already use maven, as installing and learning how to use maven is not worth the time saving of automated packaging.

At GitHub you will find an example project that you can use as a template. Just run mvn package and maven will generate a JAR file at the target directory.

References

  1. Wikipedia entry on Jython. Accessed: 2016-08-30 ↩