itom

A Measurement and Data Processing

Software Suite

2013-09-10 / 2013-09-11
What this tutorial is about

• Introduction about itom

− Why did we develop itom?
− Main features
− Python and its most important modules
− itom’s plugin system
• Show-Cases
− Macroscopic fringe projection
− Software-Plugin: GUI for GPU based ray tracer MacroSim
− Commercial confocal microscope from TWIP Optical Solutions
• Hands-on exercises
− We develop an example to calculate the offset between two images,
acquired with your webcam and create a user-developed GUI

2
Agenda

• Motivation. Why itom?

• Features
• Script Language Python
• Modular Plugin System
• The Graphical User Interface
• Licensing
• DataObject – itom’s Built-in Array Class
• Documentation and Help

4
Motivation

Matlab Labview

Data processing Easy generation of GUIs

Extensive math libraries Excellent hardware support
Integration of hardware Limited data processing and
User defined interface analysis
No unified hardware interfaces

itom
• Fast, well-established, easy to use scripting
language (Python)
• Homogeneous hardware integration
• Automation of measurement systems
• Fast data processing and analysis
• Easy to customize
5
Requirements and Solutions

Requirements Solution

Fast, performant implementation C++

Modern, user-friendly interface, independent Qt-Framework (Windows,

of hardware platform Linux, Mac OS)

Fully integrated scripting language (fast, Python (Version 3) incl. numerous

robust, easy to learn, extensive existing libraries (numpy, scipy, scikit-
libraries, well documented and supported) image, matplotlib, …)

Easy, flexible, homogenous integration of

hardware support (motors, cameras, Plugin-System
AD converter, ...) and algorithms

Using well-known, time-proven, free software OpenCV, PointCloudLibrary,

libraries where possible Qscintilla, Qwt, …

6
itom

Windows 7
7
itom
Windows 8

Debian KDE
8
itom – main features

Scripting
- Integrated Python programming environment
- Almost full Python functionality
- Controlling itom by specific Python module

Plugins
- C++ libraries (e.g. dll)
- Hardware and algorithm integration
- Integration of complex dialogs and windows

GUI
- Intuitive
- Optimized for implementation of measurement systems
- Ability to integrate customized user interfaces

9
Agenda

• Motivation. Why itom?

• Features
• Script Language Python
• Modular Plugin System
• The Graphical User Interface
• Licensing
• DataObject – itom’s Built-in Array Class
• Documentation and Help

10
Python

• Open-Source scripting language (very liberal BSD-license)

• Implemented in C
• Developed and supported since 1991
• Supports object-oriented, functional and imperative programming
paradigms
• Version 3.2 or newer supported
• Fully integrated core component of itom
• Vast number of third-party modules available for free
• Scripts are precompiled and cached for faster execution
• Integrated Python-debugger

11
Python
• Variables have an Python internal type, mainly: int, float, complex
• Casting uses the functions int(), float()…
• Assignment: a=1 a,b=1,2
• Comparison operators: ==, >, <, <=, >=, !=
• Bitwise-Operators: &, |, ~, ^
• Basic arithmetic: a = a+1, a += 1, a=a**2
• Operators also work on many non-basic types (arrays, lists, dictionaries…)

Example: Factorial

function ret = factorial(x) int factorial(int x) def factorial(x):

if(x > 1) { if (x > 1):
ret = x * factorial(x-1); if (x > 1) { return x * factorial(x-1)
else return x * factorial(x-1); else:
ret = 1; } else { return 1
end return 1;
end }
}
12
Python - Packages
• Python is embedded in itom
• itom can be controlled by Python via Numpy
itom-module numeric library
• Python is extendable by packages

Scipy
scientific library

itom

scikit-image
Matplotlib
image processing
plots and graphs
tools

13
Python-Module itom

„The bridge between Python and itom“

>> from itom import * <<

• Add menus and toolbars to itom GUI and

connect them with Python methods
• Plots arrays/matrices and camera live images

• Control hardware plugins (dataIO, actuator)

• Call algorithms from software plugins
• Online help for plugins

• Build GUIs at runtime with WYSIWYG design tool

• Connect widget’s signals to python methods
• Change properties of widgets by script commands

14
Numpy
Numeric package • Array creating and
• Support of large, multi-dimensional arrays manipulation
• Binary operations
• Large library of mathematical functions and
• Linear algebra
operators • Masked arrays
• itom’s own array object is compatible to • Polynomials
Numpy arrays. • Random Sampling
• Sorting, Searching,
Counting
Example: Solve Ax=b • Fourier Transforms
from numpy import * • …
from numpy.linalg import solve

# The system of equations we want to solve for (x0,x1,x2):

# 3 * x0 + 1 * x1 + 5 * x2 = 6
# 1 * x0 + 8 * x2 = 7
# 2 * x0 + 1 * x1 + 4 * x2 = 8

a = array([[3,1,5],[1,0,8],[2,1,4]])
b = array([6,7,8])
x = solve(a,b)
print(x) # This is our solution
[-3.28571429 9.42857143 1.28571429]

15
Scipy
Scientific Algorithms • Optimization
• Extension for numpy • Linear Algebra
• Integration
• Provide more functions from the field of
• Interpolation
numeric, statistic and optimization • FFT
• Itself extendable by scikits • Signal Processing
• ODE Solvers
• Optimization
• Basic image processing
Example: • Sparse Matrices
Find root of 2 0 around 0.3
import numpy as np
from scipy.optimize import root

def func(x):
return x + 2 * np.cos(x)

sol = np.root(func, 0.3)

sol.x
>>> array([-1.02986653])
sol.fun
>>> array([ -6.66133815e-16])

16
Matplotlib
Plotting package
• Python package for math plots
• Based on numpy
• Syntax close to Matlab
• Export in various image formats:
png, pdf, eps…
• Fully integrated in itom
• Can be integrated in custom GUIs

17
scikit-image
Image processing package • Segmentation
• Transformation
• Based on Numpy arrays
• Morphology
• Algorithms written in Python and C • Measure
• Uses Matplotlib for plotting results • IO
• Image filtering
• Rank filters
• Feature detection

Example: Entropy determination

18
Agenda

• Motivation. Why itom?

• Features
• Script Language Python
• Modular Plugin System
• The Graphical User Interface
• Licensing
• DataObject – itom’s Built-in Array Class
• Documentation and Help

19
itom Plugin System

Plugins extend the basic functionalities of itom. Each plugin is a

C++ library (.dll, .so)
Every Plugin implements one of three basic interface classes
(DataIO, Actuator, Algorithm)
Plugins (e.g. camera, motor stages…) can be instantiated from
Python or directly through the itom GUI

DataIO Actuator Algorithm

• Cameras • Motors • Algorithms

• A/D-Converters • Multi-Axes • Data Filters
• Serial Bus Machines • Complex GUIs

20
Interface “dataIO + Grabber”

Primary functionality

• getParam(..) read a parameter

• setParam(..) set a parameter
• startDevice() start camera
• stopDevice() stop camera
• acquire() take a picture
• getVal(..) / copyVal(..) load
image from camera into
itom/Python
• …

Live images from the camera can be displayed in separate windows or

integrated into custom GUIs
21
Interface “actuator”

Primary Functionality

• getParam(..) read Parameter

• setParam(..) set Parameter
• getStatus(..) get status per axis
• getPos(..) read current position
• setPosAbs/Rel() move to position
• …

Signals about position and status of the actuator can be linked to and
processed by the GUI.

22
Interface “algo”

‚Algo‘ plugins define

• Numerical algorithms
• GUI elements

Call:
• From a Python script
• By other plugins

Each method is defined by :

• Mandatory parameters (Type,
description…)
• Optional parameters
• Return values

23
Agenda

• Motivation. Why itom?

• Features
• Script Language Python
• Modular Plugin System
• The Graphical User Interface
• Licensing
• DataObject – itom’s Built-in Array Class
• Documentation and Help

24
GUI

25
GUI – Command Line

• Input/Output window for Python (Information, Warnings, Error)

• Direct execution of Python commands
• Functionality very similar to Matlab
• Auto completion of commands
• Syntax help and highlighting
26
GUI - Workspace

• Global Variables: contains all globally defined variables

• Local Variables: all local variables within a function (Debug only)
• Direct import and export to / from the Python workspace

27
GUI – File System

• Access and administration of all scripts

and files that can be opened in itom
• The default main directory is the current
working directory (similar to Matlab)
• Double click on a .py Python script will
open it in the scripting window
• Double click on supported file types will
load them into workspace

28
GUI – Plugins

• List of all available Plugins

• Sorted by category

• Allows direct instantiation of hardware

plugins

29
Scripting window

• Editor for Python scripts

• Syntax help and highlighting
• Auto completion

• Standard editor functionality

• Tabbing of multiple scripts
• Dockable into the main GUI
• Executes Scripts
• Full debugging functionality

30
Syntax Help and Auto Completion

• Auto completion Set various syntax-files (for

(selection item with tab-key) important Python modules)
in itom’s property editor in
order to enable these
features.

• Syntax help

31
Plots

• 1D, 2D, 2.5D plots

• Custom windows can be implemented
• Displayed in
− A separate window
− Docked into the main GUI
− Integrated into a custom GUI

dependent 1D-line plot

32
Custom GUIs (Qt Designer)
• Design of custom GUIs in the external Qt Designer
WYSIWYG tool (drag&drop).
• Events created by the GUI (button click) can be
linked to Python functions

Preview
Elements custom
GUI:
• Hierarchy
• Layouts

Library of Properties of each

elements element:
Can be adapted by
Python scripts in
itom Designer- itom
Plugins 33
Custom GUIs (Qt Designer)
Dialog design with Qt Designer:

Dialog: gui

Script logic with python:

gui.btnOk[“text”] = “OK”
1. Access properties gui.comboOs.call(“addItems”, [“Windows”,”Linux”])

2. Connect signals with def clickMe():

print(“operating system”, gui.comboOs[“currentText”])
Python methods
gui.btnOk.connect(“clicked()”, clickMe)

34
Multithreading

Main Thread Plugin Thread

All GUI Elements Asynchronous

• Main Window Communication
Each plugin runs
• Scripting using Qt-
in its own,
Window Signal/Slots
separate thread.
• Plots
The corresponding
GUI elements are
Script Organizer
pushed to the
Python Thread main thread.
UI Organizer

Python Interpreter
Plugin Organizer

Python Debugger
…

35
Agenda

• Motivation. Why itom?

• Features
• Script Language Python
• Modular Plugin System
• The Graphical User Interface
• Licensing
• DataObject – itom’s Built-in Array Class
• Documentation and Help

36
License

• itom (main application) is “Open Source” (LGPL)

• itom-SDK (resources common to the main application and plugins)

are distributed under the LGPL-licence + itom-exception. The itom
exception allow the inclusion and linking of additional components
independent of those components licensing against all data included
in the SDK.

• Plugins can be subject to any (including proprietary) licenses. The

ITO offers a number of generic plugins under the LGPL.

• Designer-Plugins (plots…), similarly, can be subject

to any licenses.

37
Agenda

• Motivation. Why itom?

• Features
• Script Language Python
• Modular Plugin System
• The Graphical User Interface
• Licensing
• DataObject – itom’s Built-in Array Class
• Documentation and Help

38
Data Object
Goal:
• Different basic types of data (including complex)
• Processing of large, multi-dimensional data sets (series of images)
• Compatible with Matlab, Numpy, OpenCV

Implementation:
• DataObject very similar to OpenCV data structures
• Basic data types supported: int8, uint8, int16, uint16, int32, uint32, float,
double, complex(float), complex(double)
• DataObject supports tags (axes units, descriptions, title…)

39
Data Storage
Series of 2D-images

3D data stack

Plane 0
Plane 1
Plane 2

40
Data Storage
Assume: Series of 2D-images (3 x 2 x 5)
(0,0,0)

Plane 0
Plane 1
Plane 2
(2,1,0) (2,1,4)

C / Matlab: continuous chunk of memory

(0,0,0) (0,1,0) (1,0,0) (2,1,4)

Plane 0 Plane 1 Plane 2

+ Uniform, quick and easy access to multi-dimensional arrays

– Memory allocation error for „big“ arrays

41
Data Storage
DataObject:

0 (0,0,0) (0,1,0) Plane 0

1 (1,0,0) Plane 1

2 (2,1,4) Plane 2

+ Less allocation errors due to distributed chunks of memory

– Slightly more complex access to memory

DataObject (continuous): Compatibility to C-style arrays

0 1 2

(0,0,0) (0,1,0) (1,0,0) (2,1,4)

Plane 0 Plane 1 Plane 2

42
Agenda

• Motivation. Why itom?

• Features
• Script Language Python
• Modular Plugin System
• The Graphical User Interface
• Licensing
• DataObject – itom’s Built-in Array Class
• Documentation and Help

43
User Documentation

• User documentation displayed with Qt Assistant

• Can be exported to pdf, html…

Help pages

• Table of content
• Search
• Bookmarks

itom.bitbucket.org/latest/docs

44
Additional User Help within Python
1. Syntax help and auto completion in
the Python editor

2. Customizable, context sensitive

syntax highlighting

3. Python-internal help system using the

command help(…)

4. Additional information and help about

available plugins or algorithms using
the commands
pluginHelp(…),
filterHelp(…),
widgetHelp(…)
45
itom

Questions?

46
itom
Show Cases
Show-Case I: Fringe Projection
Situation

A flexible fringe projection setup

(structured light) for student
projects and public presentation is
been developed

Objective

Provide a GUI for such a system

to demonstrate the function

Allow students to run batch

processes for system
characterization

Provide flexibility to change between several

evaluation or calibration methods and hardware
components.
48
Show-Case I: Triangulation
Laser

camera
chip

H ∆x

camera lens

∆z
object

49
Show-Case I: Structured Illumination
Cosine-fringes (mod 2PI)
LCOS-Display or
DMD-Projector
absolute coding
Grey-Code

50
Show-Case I: Basic Set-up

Camera + lens

Projector

51
Show-Case II: MacroSim

Situation

An open source GPU based ray-tracing tool has been

developed at ITO

The native tool is command-line based

Objective

Provide a GUI for MacroSim in order to simplify the

creation of new scenes and execute simulations

For the future it should be possible to run both the real

setup and its corresponding simulation with the same tool.

52
Show-Case II: MacroSim

Solution

Create an itom software plugin that provides its own GUI

and communicates with the tool MacroSim

itom MacroSim
MacroSim Tracer
Plugin

MacroSim can use functionalities contained in itom

Tracer can also be started by Python
Batch execution possible using appropriate Python script
Results of tracer are available in itom

53
Raytracing: A versatile tool

Raytracing is perfectly linear

→ Raytracing is perfectly parallelizable

Modell adaptiert von US Patent Publication US 6522484 B1, K-H

Schuster, Carl-Zeiss-Stiftung, (1999) 54
Parallelization of Raytracing

CPU-Parallelization
- very flexible
- straightforward implementation
- More than 4 cores quickly become expensive

GPU-Parallelization
- Restriction to Thread Coherence
- Specific Implementation
- Standard GPUs come with 200-500 cores

Images taken from CUDA programming guide 3.2 56

Parallelization of Raytracing

GPU-Parallelization
- Restriction to Thread Coherence
- Specific Implementation
- Standard GPUs come with 200-500 cores

GPU accelerated Tool: MacroSim

- Based on nVidia® OptiX™ acceleration engine
- Plugin to ITOs itom software
- imports glass catalog from Zemax®
- Published under GPL at https://bitbucket.org/itom/macrosim
- „An open source GPU-accelerated ray tracer for optical simulation“,
submitted for publication to Optical Engineering.

Images taken from CUDA programming guide 3.2 57

Parallelization of Raytracing
MacroSim Plugins contains one GUI and
some callable functions

58
Interaction with itom

GUI based
• Start MacroSim GUI by Python command (createNewPluginWidget)
• Start simulation manually
• GUI emits a signal with the final detector matrix (dataObject)
• Connect a Python function to this signal (called when simulation done)

Script based
• Optional: Start MacroSim GUI and create scene (XML-file)
• Call function runSimulation of MacroSim plugin and pass XML-file
(simulation is executed)
• The function finally returns the detector matrix as dataObject

59
Show-Case III: Confocal Microscopy

60
Show-Case III: Confocal Microscopy

Situation

A confocal microscope is being developed by Twip Os

(spin-off of ITO)

Objective
itom should be used to…
− control the measurement process
− provide a user-friendly control panel
− visualize the results
− provide functionality for data evaluation (roughness,
alignment, geometrical fitting…)

61
Show-Case III: Confocal Microscopy

Object

FWHM ~ 1/NA

Point Detector
62
Show-Case III: Confocal Microscopy
Object

Point Detector

63
Show-Case III: Confocal Microscopy
GUI

64