Deconvolution: infrastructure and reference algorithms
INRA
| Please use this identifier to cite or link to this publication: http://hdl.handle.net/10380/3207 |
Published in The Insight Journal - 2010 July-December.
Submitted by Gaetan Lehmann on 08-30-2010.
The deconvolution, also called deblurring, tries to revert the optical distortion introduced during the aquisition of the image. It is a family of image processing which can be classed in the larger family of image restoration.
Deconvolution is a very difficult problem, and many algorithms have been proposed to solve it, with different strenght and weakness which may depend on the context where they are used. As a consequence, it is desirable to have several algorithms available when trying to restore some images. The different algorithms are often built on a similar principle, making possible to share a large part of their API in their implementation. Also, the most generic operations related to deconvolution should be reusable in order to avoid code duplication and ease the implementation of new algorithms.
In this contribution, the infrastructure for the implementation of several deconvolution algorithms is proposed. Based on this infrastructure, twelve simple deconvolution algorithms of reference are also provided.
Deconvolution is a very difficult problem, and many algorithms have been proposed to solve it, with different strenght and weakness which may depend on the context where they are used. As a consequence, it is desirable to have several algorithms available when trying to restore some images. The different algorithms are often built on a similar principle, making possible to share a large part of their API in their implementation. Also, the most generic operations related to deconvolution should be reusable in order to avoid code duplication and ease the implementation of new algorithms.
In this contribution, the infrastructure for the implementation of several deconvolution algorithms is proposed. Based on this infrastructure, twelve simple deconvolution algorithms of reference are also provided.
Code
Reviews
A very important contribution for microscopy applications
by Cory Quammen on 2010-10-07 21:16:36 for revision #1 
expertise: 4 sensitivity: 5 
Summary:
Corrupting images by simulated noise is an essential process for generating realistic simulated images where the ground truth data is known and corrupted by noise with known characteristics.
Hypothesis:Deconvolution algorithms can be added to ITK in a way that uses common API elements to minimize code duplication.
Evidence:The author provides an implementation of numerous deconvolution algorithms organized into a design that minimizes code duplication. The class hierarchy also serves as a nice conceptual taxonomy of deconvolution algorithms.
Open Science:All source code and example images are included with the publication.
Reproducibility:I could not compile this contribution on Windows because the type uint32 is not defined.
Use of Open Source Software:All the software provided is open source and extends the open source ITK library.
Open source Contributions:It took just a few short minutes to configure, compile, and run the examples.
Code Quality :The code is high quality and appears to conform to ITK's coding style by visual inspection.
Quality of the data :
The test images were useful for demonstrating the operation of the noise filters.
Interest:Adding simulated noise is incredibly useful for many, many fields.
Free comment :This is a fantastic contribution.
Quick Comments
Resources
|
|
| Download Package | |
| Download Paper, View Paper | |
| Download Source code | |
| Github | |
Statistics more
|
|
| Global rating: | 
|
| Review rating: | [review]
|
| Code rating: | |
| Paper Quality: |
|
Information more
|
|
| Categories: | Filtering, Programming |
| Keywords: | deconvolution, fft, |
| Toolkit: | CMake, ITK |
| Export citation: | |
Share
|
View license
Loading license...
Send a message to the author
