https://github.com/psteinb/libmultiviewnative

Native Implementation (CPU & GPU) of the Iterative Bayesian Deconvolution Algorithm in the Fiji Multi-View Deconvolution Plugin by Stephan Preibisch
https://github.com/psteinb/libmultiviewnative

Last synced: 7 days ago
JSON representation

Native Implementation (CPU & GPU) of the Iterative Bayesian Deconvolution Algorithm in the Fiji Multi-View Deconvolution Plugin by Stephan Preibisch

• Host: GitHub
• URL: https://github.com/psteinb/libmultiviewnative
• Owner: psteinb
• License: gpl-2.0
• Created: 2014-04-11T12:36:10.000Z (almost 11 years ago)
• Default Branch: master
• Last Pushed: 2016-02-17T10:40:56.000Z (almost 9 years ago)
• Last Synced: 2023-10-26T10:06:35.853Z (about 1 year ago)
• Language: C++
• Size: 475 KB
• Stars: 8
• Watchers: 5
• Forks: 3
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# GPU based Multi-View Deconvolution using cuFFT

This package aims at providing an easy-to-use API for calling GPU deconvolution routines from the Fiji/SPIM_Registration plugin ( https://github.com/fiji/spimreconstruction ). It is written in C++ and offers a C API of plain methods and C structs to call from any client binary or library. The implementation is based on .

# Building the Library

## Operating Systems

We currently support the following operating systems:

* OSX 10.8 or higher (experimental)

* Fedora Linux 19 or higher

* CentOS Linux 6.3 or higher

## Dependencies

* CUDA 5 or later (the nvcc compiler, libcudart and libcufft) must be available.

* cmake versions later than 2.8 ( FindCUDA.cmake sources are needed )

* boost 1.42 or later

* fftw 3.1 or later

## Install:

To compile the shared library libMultiViewNative.so, invoke

```bash

$ mkdir build

$ cd build

```

If a specific installation destination is to be used

```bash

$ cmake -DCMAKE_INSTALL_PREFIX=/path/to/lib ..

$ make

$ make install

```

or else (will install library in system default library paths, administrator rights required)

```bash

$ cmake .. 

$ make

$ make install

```

# Using the Library

*Note: The GPU is unstable and not yet production ready. The CPU API can be used!*

## Fiji related

In order to call the GPU methods from Java, simply create a directory called lib/linux64 under the fiji root directory. Create a symlink to libMultiViewNative.so (compiled for your architecture) here.

## License

see LICENSE

## Contact

For questions, bug reports, improvement suggestions or else, use the github issue tracker of this repo.

## Known Issues

On OS X, CUDA is (at the time of writing) INCOMPATIBLE with the installed llvm based compilers used by default (e.g. /usr/bin/c++). To resolve this, the installation procedure uses an llvm based gcc version for compiling and testing.

## Disclaimer

The library at hand is an evolving project. It is designed as the native implementation of http://fiji.sc/Multi-View_Deconvolution (cf. ). The ultimate goal is to implement the performance sensitive parts of the Multi-View_Deconvolution in this library which can provide significant speed-ups:  (recording of the talk),  (slides)