DIPlib  3.0.alpha A Library for Quantitative Image Analysis

## Introduction

The purpose of the DIPlib project is to provide a one-stop library and development environment for quantitative image analysis, be it applied to microscopy, radiology, astronomy, or anything in between.

There are other image processing/analysis libraries available, some of them hugely popular. Why do we keep investing time in developing and improving DIPlib? The short answer is that we believe DIPlib offers things that are not available elsewhere. The library is built on the following three principles:

1. Precision:

We implement the most precise known methods, and output often defaults to floating-point samples. The purpose of these algorithms is quantification, not approximation.

2. Ease of use

We use modern C++ features to provide a simple and intuitive interface to algorithms, with expressive syntax, default values, and little boiler-plate code required from the user. There is no need to be aware of an image's data type to use the algorithms effectively.

Furthermore, developing an image analysis program involves a lot of trial-and-error, rapid prototyping approaches are applicable: the edit-compile-run loop should be quick. We aim for short compile times with pre-compiled algorithms and few public templates.

3. Efficiency

We implement the most efficient known algorithms, as long as they don't compromise precision. Ease-of-use features might also incur a slight overhead in execution times. The library can be used in high-throughput quantitative analysis pipelines, but is not designed for real-time video processing.

Algorithms in DIPlib typically accept input images of any data type (though, of course, some algorithms are specific to binary images, or cannot handle complex images, etc.) and any number of dimensions (algorithms that are limited to one specific dimensionality typically show so in their name). The image data type and dimensionality do not need to be known at compile time. Images can have pixels that are vectors or matrices, for some examples on how this relates to the three points above, see Why tensors?.

There are many other unique things about DIPlib, we encourage you to explore the documentation to learn more about it. A good place to start are the following documentation pages:

## Interfaces and bindings

Currently, DIPlib 3 has interfaces or bindings to the following packages:

• MATLAB: DIPimage 3 is a MATLAB toolbox that gives access to most functionality in DIPlib, but goes beyond that by providing a lot of additional functionality as M-file functions.
• Python: PyDIP 3 is a thin wrapper of most functionality in DIPlib.
• OpenCV: dip_opencv provides copyless conversion to and from OpenCV images, for OpenCV version 2 and newer.

## Support

The DIPlib project was originally developed at the Pattern Recognition Group of Delt University of Technology, in the Netherlands. DIPlib 3 is being developed primarily by volunteers, but has had some financial support from:

• A European Research Council grant to Bernd Rieger, TU Delft
• Flagship Biosciences, Inc.

See A short history for a list of contributors.

## Contributing

If you want to contribute to the DIPlib project, there are many different ways of doing so:

• Help port algorithms from DIPlib 2 to the new infrastructure. Please coordinate with Cris Luengo before you get started. He can share the old code for the algorithms that need porting. See Work plan for DIPlib 3 for a list of stuff to do.
• Write new algorithms. If you have an algorithm that you'd like to contribute to the project, we'll be happy to see it!
• Create an interface to another library or scripting language.
• Create tutorials for how to use DIPlib, DIPimage and/or PyDIP.
• Fix bugs or improve documentation.
• Add code to the unit tests.
• Create a nice Doxygen theme for the documentation, or create a logo for the project.

In any of these cases, see CONTRIBUTING.md to learn how to make optimal use of your time. Don't be offended if you receive requests for modifications before your work is merged with the project.

Your documentation and code contributions will carry the same licencing terms as the rest of the library, you keep the copyright to any substantial contribution.

Copyright 2014-2018 Cris Luengo
Copyright 1995-2014 Delft University of Technology

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at