Gimp Plugin using CUDA: Multiresolution Gradient Adaptive Filter
Cuda C++ CMake
Switch branches/tags
Nothing to show
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
doc
AUTHORS
CMakeLists.txt
COPYING
INSTALL
README
defines_cpu.hpp
defines_gpu.hpp
gimp_gui.cpp
gimp_gui.hpp
gimp_main.cpp
gimp_main.hpp
multi_res_cpu.cpp
multi_res_cpu.hpp
multi_res_device.cu
multi_res_host.cu
multi_res_host.hpp

README

Gimp Plugin: Multiresolution gradient adaptive filter
=====================================================

The gimp plugin implements a multiresolution gradient adaptive filter, as
described in [1], however the filter kernel was replaced by a bilateral filter,
as described in [2]. The major advantage of this filter is the fact that the
filter preserves edges while noise is removed.
The multiresolution filter consists of 3 major steps (see also doc/mrfilt.eps):

1. Decompose: the gaussian (g0) and laplace (l0) pyramid are constructed
2. Filter: the bilateral filter (f0) gets applied to the laplace pyramid
3. Reconstruct: the output image (r0) is reconstructed from the filted data

The multiresolution filter operates on gray-scale images and requires the image
resolution to be a power of 2. In case that the image doesn't fulfill this
requirement, the image is padded with mirrored pixel values. The
multiresolution filter works on 5 stages and the parameters for the bilateral
filter kernel can be adjusted (see also: INSTALL file).

---

[1] D. Kunz, K. Eck, H. Fillbrandt, and T. Aach. Nonlinear multiresolution
gradient adaptive filter for medical images. Proceedings of SPIE, 5032:732,
2003.

[2] C. Tomasi and R. Manduchi. Bilateral filtering for gray and color images.
Proceedings of the Sixth International Conference on Computer Vision, page
839, 1998.