December 2012 - Olena-patches

[Olena] #147: Write a script to automatically generate file skeletons for Olena
by Olena 29 Aug '13

29 Aug '13

#147: Write a script to automatically generate file skeletons for Olena -----------------------+---------------------------------------------------- Reporter: levill_r | Owner: Olena Team Type: task | Status: new Priority: trivial | Milestone: Component: Milena | Version: 1.0 Keywords: generator | -----------------------+---------------------------------------------------- A Ruby script would be nice, but the language doesn't really matter. -- Ticket URL: <https://trac.lrde.org/olena/ticket/147> Olena <http://olena.lrde.epita.fr> Olena, a generic and efficient C++ image processing library.

2 4

[Olena] #190: Handle Trimesh's dependencies on third-party packages.
by Olena Trac 29 Aug '13

29 Aug '13

#190: Handle Trimesh's dependencies on third-party packages. --------------------------------+------------------------------------------- Reporter: levill_r | Owner: levill_r Type: task | Status: new Priority: minor | Milestone: Olena 1.0 Component: Olena | Version: 1.0 Keywords: Autoconf configure | --------------------------------+------------------------------------------- * OpenGL/Mesa (for Trimesh) * We should check for the presence of `GL/gl.h`, `GL/glu.h` and `libXmu.{a,so}` to enable OpenGL conditionally. * Note: on Debian, Mesa users have to install these packages: * `mesa-common-dev` (for `GL/gl.h`) * `libglu1-mesa-dev` (for `GL/glu.h`) * `libxmu-dev` (for `libXmu.{a,so}`). -- Ticket URL: <https://trac.lrde.org/olena/ticket/190> Olena <http://olena.lrde.epita.fr> Olena, a generic and efficient C++ image processing library.

1 2

[Olena] #238: Have Swilena be compatible with (also) Python 2.6 (and greater)
by Olena Trac 29 Aug '13

29 Aug '13

#238: Have Swilena be compatible with (also) Python 2.6 (and greater) -------------------------+-------------------------------------------------- Reporter: levill_r | Owner: Olena Team Type: enhancement | Status: new Priority: major | Milestone: Component: Swilena | Version: 1.0 Keywords: | -------------------------+-------------------------------------------------- Swilena does not work with Python 2.6 (and probably greater versions). Swig seems to generates wrappers correctly, and they compile fine, but tests fail at run time. There may have been changes in Python's API between versions 2.5 and 2.6 (perhaps iterators?). Investigate. -- Ticket URL: <https://trac.lrde.org/olena/ticket/238> Olena <http://olena.lrde.epita.fr> Olena, a generic and efficient C++ image processing library.

1 1

[Olena] #160: Revamp complexes
by Olena 29 Aug '13

29 Aug '13

#160: Revamp complexes ----------------------------+----------------------------------------------- Reporter: levill_r | Owner: Olena Team Type: enhancement | Status: new Priority: minor | Milestone: Olena 1.0 Component: Milena | Version: 1.0 Keywords: cleanup rename | ----------------------------+----------------------------------------------- * Move complexes from `mln/core/` to `mln/topo/`. * The '''dynamic''' access(es) to the set of ''n''-faces of a complex (e.g., `mln::complex::nfaces(unsigned)`) has (have) a linear complexity! We should get rid of that. We could create an index at the creation of the complex, mapping dynamic dimension numbers to the corresponding (static) mixin/data. Ask Roland for details. -- Ticket URL: <https://trac.lrde.org/olena/ticket/160> Olena <http://olena.lrde.epita.fr> Olena, a generic and efficient C++ image processing library.

2 12

[Olena] #227: List of dependencies out of date in the wiki
by Olena Trac 29 Aug '13

29 Aug '13

#227: List of dependencies out of date in the wiki --------------------------+------------------------------------------------- Reporter: jonathan | Owner: Olena Team Type: defect | Status: new Priority: major | Milestone: Component: Milena | Version: 1.0 Keywords: Dependencies | --------------------------+------------------------------------------------- In the download page, [http://www.lrde.epita.fr/cgi- bin/twiki/view/Olena/Download], the list of dependencies is out of date. Especially for the git repository version, additional dependencies are needed to build the documentation. It should be clarified in the wiki or the building procedure should be improved. ****************** A non exhaustive(?) list of dependencies: autogen automake1.7 autoconf make g++ libboost-dev texlive-latex-base texlive-extra-utils pgf HeVeA latex2html doxygen imagemagick dot2tex -- Ticket URL: <https://trac.lrde.org/olena/ticket/227> Olena <http://olena.lrde.epita.fr> Olena, a generic and efficient C++ image processing library.

1 2

[Olena] #232: Update contributors' list on Olena's Web site
by Olena Trac 25 Jul '13

25 Jul '13

#232: Update contributors' list on Olena's Web site ----------------------+----------------------------------------------------- Reporter: levill_r | Owner: Olena Team Type: task | Status: new Priority: major | Milestone: Olena 1.1 Component: Web | Version: 1.0 Keywords: | ----------------------+----------------------------------------------------- This page : https://www.lrde.epita.fr/cgi-bin/twiki/view/Olena/Contributors should be updated with the latest version (e.g., from branch `next`) of this file : source:AUTHORS Better: write a script to automate the conversion, so as to simplify the process of updating this page down to a copy-and-paste operation. -- Ticket URL: <https://trac.lrde.org/olena/ticket/232> Olena <http://olena.lrde.epita.fr> Olena, a generic and efficient C++ image processing library.

1 2

[Olena] #236: Implement a VTK loader
by Olena Trac 12 Feb '13

12 Feb '13

#236: Implement a VTK loader ----------------------+----------------------------------------------------- Reporter: levill_r | Owner: Olena Team Type: task | Status: new Priority: minor | Milestone: Olena 1.1 Component: Milena | Version: 1.0 Keywords: I/O | ----------------------+----------------------------------------------------- I.e., `mln::io::vtk::load`. See also source:milena/mln/io/vtk/save.hh. -- Ticket URL: <https://trac.lrde.org/olena/ticket/236> Olena <http://olena.lrde.epita.fr> Olena, a generic and efficient C++ image processing library.

1 1

[Olena] #188: Implement more windows, neighborhoods, and corresponding iterators for complex-based images
by Olena Trac 12 Feb '13

12 Feb '13

#188: Implement more windows, neighborhoods, and corresponding iterators for complex-based images ----------------------+----------------------------------------------------- Reporter: levill_r | Owner: levill_r Type: task | Status: new Priority: minor | Milestone: Olena 1.1 Component: Milena | Version: 1.0 Keywords: | ----------------------+----------------------------------------------------- This ticket is a sequel to #162. In addition to the neighborhoods, windows, and iterators for complex-based images from #162, implement windows/neighborhoods (and corresponding iterators) returning the following elements (where ''n'' is a (fixed) dimension and ''p'' a psite on an ''n''-face). `complex_cell_window`:: the set of the faces in the « cell » including ''p'' (named « ''p''-hat » in couprie.08.pami), i.e. the set of all ''m''-faces transitively adjacent to ''p'', where ''m'' is in [0, ''n''-1].[[BR]] In that definition, ''p'' is said adjacent to an ''m''-face ''q'' if there is a sequence (''m,,1,,'', ''m,,2,,'', ..., ''m,,k,,'') of faces so that * ''m,,1,,'' is an (''n''-1)-face adjacent to ''p'' ; * ''m,,2,,'' is an (''n''-2)-face adjacent to ''m,,1,,'' ; * ... (and so on) * ''m,,k,,'' is an (''m''+1)-face adjacent to ''q''. `complex_boundary_cell_window`:: Likewise, but excluding ''p'' (named « ''p''-hat* » in couprie.08.pami). And maybe: `complex_lower_dims_connected_window`:: the set of ''n''-faces sharing a (''n''-1)-face or (''n''-2)-face etc. (by transitivity) with ''p'' (is it useful?); `complex_higher_dims_connected_window`:: the set of ''n''-faces sharing a (''n''+1)-face or (''n''+2)-face etc. (by transitivity) with ''p'' (is it useful?); * what else? As in #139, * we might want to factor things using implementation classes: * ``mln::internal::complex_vicinity`` * ``mln::internal::complex_vicinity_piter`` (we might even be able to factor them with graph-based ones); * we could have one or several generic classes, using static or dynamic predicates, to implement those windows and neighborhoods. Note that some of these iterators might implement operators on complexes, see http://en.wikipedia.org/wiki/Simplicial_complex#Closure.2C_star.2C_and_link. Do not forget to update [wiki:Olena/ComplexBasedImages]. -- Ticket URL: <https://trac.lrde.org/olena/ticket/188> Olena <http://olena.lrde.epita.fr> Olena, a generic and efficient C++ image processing library.

1 1

branch exp/inpainting updated: olena-2.0-35-g33248b9
by git＠lrde.epita.fr 20 Dec '12

20 Dec '12

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "Olena, a generic and efficient image processing platform". The branch exp/inpainting has been updated via 33248b9f4649130b71e97d3a15599a52502d9643 (commit) from 2f5a09b9567b2d56843c93952104d5cb26774ce2 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- 33248b9 backup ----------------------------------------------------------------------- Summary of changes: milena/mln/inpainting/derivative_kernels.hh | 34 +++-- milena/mln/inpainting/fast_isotropic_diffusion.hh | 2 - milena/mln/inpainting/lpde.hh | 64 +++++++--- milena/mln/inpainting/mean_diffusion.hh | 133 +++++++++++++++++++++ milena/mln/inpainting/tv.hh | 93 ++++++++++++++ 5 files changed, 292 insertions(+), 34 deletions(-) create mode 100644 milena/mln/inpainting/mean_diffusion.hh create mode 100644 milena/mln/inpainting/tv.hh hooks/post-receive -- Olena, a generic and efficient image processing platform

1 0

olena-2.0-35-g33248b9 backup
by Coddy Levi 20 Dec '12

20 Dec '12

--- milena/mln/inpainting/derivative_kernels.hh | 34 +++-- milena/mln/inpainting/fast_isotropic_diffusion.hh | 2 - milena/mln/inpainting/lpde.hh | 64 +++++++--- milena/mln/inpainting/mean_diffusion.hh | 133 +++++++++++++++++++++ milena/mln/inpainting/tv.hh | 93 ++++++++++++++ 5 files changed, 292 insertions(+), 34 deletions(-) create mode 100644 milena/mln/inpainting/mean_diffusion.hh create mode 100644 milena/mln/inpainting/tv.hh diff --git a/milena/mln/inpainting/derivative_kernels.hh b/milena/mln/inpainting/derivative_kernels.hh index 5b642ee..a7f3c34 100644 --- a/milena/mln/inpainting/derivative_kernels.hh +++ b/milena/mln/inpainting/derivative_kernels.hh @@ -2,23 +2,29 @@ namespace mln { namespace inpainting { - template <typename T> - const T* make_dx_kernel() + namespace { - static const T ws[] = { 0.1, 0.2, 0.1, - 0, 0, 0, - -0.1, -0.2, -0.1 }; - return ws; - } + const float a = 61; + const float b = 17; - template <typename T> - const T* make_dy_kernel() - { - static const T ws[] = { 0.1, 0, -0.1, - 0.2, 0, -0.2, - 0.1, 0, -0.1 }; + template <typename T> + const T* make_dx_kernel() + { + static const T ws[] = { b, a, b, + 0, 0, 0, + -b, -a, -b }; + return ws; + } + + template <typename T> + const T* make_dy_kernel() + { + static const T ws[] = { b, 0, -b, + a, 0, -a, + b, 0, -b }; - return ws; + return ws; + } } template <typename T> diff --git a/milena/mln/inpainting/fast_isotropic_diffusion.hh b/milena/mln/inpainting/fast_isotropic_diffusion.hh index d63f9d4..44a8989 100644 --- a/milena/mln/inpainting/fast_isotropic_diffusion.hh +++ b/milena/mln/inpainting/fast_isotropic_diffusion.hh @@ -76,8 +76,6 @@ namespace mln std::swap(u_, u_next_); } - - std::cout << iterations << "\t"; } } /* mln::inpainting */ diff --git a/milena/mln/inpainting/lpde.hh b/milena/mln/inpainting/lpde.hh index 5f34d83..f38b7a3 100644 --- a/milena/mln/inpainting/lpde.hh +++ b/milena/mln/inpainting/lpde.hh @@ -24,7 +24,10 @@ #include <mln/core/alias/vec2d.hh> #include <mln/inpainting/derivative_kernels.hh> + #include <mln/inpainting/fast_isotropic_diffusion.hh> +#include <mln/inpainting/mean_diffusion.hh> +#include <mln/inpainting/tv.hh> #include <mln/core/concept/accumulator.hh> @@ -42,7 +45,7 @@ namespace mln struct lpde { public: - typedef double float_type; + typedef float float_type; lpde(float dt = 1.0f, float dx = 1.0f, float t_f = 20.0); @@ -96,19 +99,30 @@ namespace mln return components_product(mln_trait_value_nature(T)(), a, b); } + template <typename T> + T normalize(T v) + { + lpde::float_type magnitude = norm::l2(v); + + magnitude = std::max((lpde::float_type)1.0, magnitude); + + return v / magnitude; + } + template <typename I, typename M> lpde::float_type diffuse(const I& u, I& u_next, - //const I& f, const std::vector<lpde::float_type>& w, const M& mask, - lpde::float_type& max_diff_norm) + lpde::float_type& max_diff_norm, + I& u_x, + I& u_y) { typedef lpde::float_type float_type; typedef mln_value(I) T; const float_type activation_threshold = 0; - + const unsigned derivative_kernels_size = 5; const float_type* derivative_kernel[derivative_kernels_size]; @@ -127,13 +141,20 @@ namespace mln mln_pixter(const I) p(u); mln_pixter(I) p_next(u_next); - //mln_pixter(const I) p_f(f); mln_pixter(const M) p_m(mask); + + mln_pixter(I) pu_x(u_x); + mln_pixter(I) pu_y(u_y); + mln_qixter(const I, const window2d) q(p, full3x3); float_type sse = 0; - - for_all_3 (p, p_next, p_m) + + for (p.start(), p_next.start(), p_m.start(), + pu_x.start(), pu_y.start(); + p.is_valid(); + p.next(), p_next.next(), p_m.next(), + pu_x.next(), pu_y.next()) { if (p_m.val()) { @@ -142,9 +163,6 @@ namespace mln if (fabs(w[1]) > activation_threshold) p_next.val() = p.val() * w[1]; - // if (fabs(w[0]) > activation_threshold) - // p_next.val() += w[0] * p_f.val(); - T p_derivative[derivative_kernels_size]; for (unsigned i = 0; i < derivative_kernels_size; ++i) @@ -172,6 +190,9 @@ namespace mln const T p_yy2 = components_product(p_yy, p_yy); const T p_xy2 = components_product(p_xy, p_xy); + pu_x.val() = normalize(p_x); + pu_y.val() = normalize(p_y); + //||∇u||^2 if (fabs(w[2]) > activation_threshold) p_next.val() += w[2] * (p_x2 + p_y2); @@ -200,6 +221,8 @@ namespace mln } } + inpainting::tv(u_next, u_x, u_y, mask, max_diff_norm); + return sse; } @@ -207,10 +230,11 @@ namespace mln void lpde::operator()(I<T>& src, const I<bool>& mask) { - //fast_isotropic_diffusion fast_iso; - //fast_iso(src, mask); - - //const I<T> f(src); + // mean_diffusion mean_diff; + // mean_diff(src, mask); + + // fast_isotropic_diffusion fast_iso; + // fast_iso(src, mask); I<T>& u_ = src; I<T> u_next_(src); @@ -229,17 +253,21 @@ namespace mln float_type min_max_diff_norm = std::numeric_limits<float_type>::max(); + I<T> u_x(u_.domain()); + I<T> u_y(u_.domain()); + float t = 0; while (t < this->t_f && max_diff_norm > CONVERGENCE_THRESHOLD - //&& sse < sse_old - && !std::isnan(sse)) + && !std::isnan(max_diff_norm) + ) { + max_diff_norm = 0; sse_old = sse; - sse = diffuse(*u, *u_next, /* f, */ w, mask, max_diff_norm); + sse = diffuse(*u, *u_next, w, mask, max_diff_norm, u_x, u_y); if (max_diff_norm < min_max_diff_norm) min_max_diff_norm = max_diff_norm; @@ -258,4 +286,4 @@ namespace mln } } /* mln::inpainting */ -} /* mln*/ +} /* mln */ diff --git a/milena/mln/inpainting/mean_diffusion.hh b/milena/mln/inpainting/mean_diffusion.hh new file mode 100644 index 0000000..e0e3f48 --- /dev/null +++ b/milena/mln/inpainting/mean_diffusion.hh @@ -0,0 +1,133 @@ +#include <mln/core/macros.hh> +#include <mln/core/concept/image.hh> +#include <mln/core/alias/point2d.hh> +#include <mln/pw/value.hh> +#include <mln/core/alias/neighb2d.hh> + +#include <mln/core/image/dmorph/image_if.hh> + +#include <mln/core/concept/function.hh> + +#include <mln/data/transform.hh> + +#include <mln/literal/colors.hh> +#include <mln/literal/zero.hh> +#include <mln/pw/image.hh> +#include <mln/convert/to_p_array.hh> +#include <mln/core/var.hh> + +namespace mln +{ + + namespace inpainting + { + + struct mean_diffusion + { + public: + typedef float float_type; + + template <template <typename> class I, typename T> + void operator()(I<T>& src, + const I<bool>& mask_); + }; + + namespace + { + template <typename I, typename N> + struct contour : public Function_v2b< contour<I,N> > + { + typedef bool result; + + contour(const I& ima, const N& neighborhood) + : ima_ (ima), + neighborhood_ (neighborhood) + { + } + + bool operator()(const mln_psite(I)& p) const + { + if (ima_(p)) + { + mln_niter(N) n(neighborhood_, p); + + for_all(n) + { + if (!ima_(n)) + return true; + } + } + + return false; + } + + private: + const I& ima_; + const N& neighborhood_; + }; + + template <template <typename> class I, typename T> + void diffuse(I<T>& src, + const p_array<mln_psite(I<bool>)>& ctr, + I<bool>& mask, + const neighb2d& neighborhood) + { + + mln_piter(p_array<mln_psite(I<bool>)>) p(ctr); + mln_niter_(neighb2d) n(neighborhood, p); + + for_all(p) + { + T accu = literal::zero; + + unsigned nb = 0; + + for_all(n) + { + if (!mask(n)) + { + accu += src(n); + ++nb; + } + } + + src(p) = accu / nb; + } + } + + } /* !mln::inpainting::{anonymous} */ + + template <template <typename> class I, typename T> + void mean_diffusion::operator()(I<T>& src, + const I<bool>& mask_) + { + const neighb2d neighborhood = c8(); + + I<bool> mask(mask_.domain()); + data::paste(mask_, mask); + + unsigned contour_nsites = std::numeric_limits<unsigned>::max(); + + while (contour_nsites > 0) + { + contour<I<bool>, neighb2d> ctr = contour<I<bool>, neighb2d>(mask, neighborhood); + p_array<mln_psite(I<bool>)> ctr_set = convert::to_p_array(mask | ctr); + contour_nsites = ctr_set.nsites(); + + if (contour_nsites > 0) + { + diffuse(src, ctr_set, mask, neighborhood); + + mln_piter(p_array<mln_psite(I<bool>)>) p(ctr_set); + for_all (p) + { + mask(p) = false; + } + } + } + } + + } /* mln::inpainting */ + +} /* mln */ + diff --git a/milena/mln/inpainting/tv.hh b/milena/mln/inpainting/tv.hh new file mode 100644 index 0000000..d5bed23 --- /dev/null +++ b/milena/mln/inpainting/tv.hh @@ -0,0 +1,93 @@ +#include <mln/core/alias/window2d.hh> +#include <mln/core/concept/iterator.hh> +#include <mln/algebra/vec.hh> +#include <mln/convert/from_to.hh> +#include <mln/norm/l2.hh> + +namespace mln +{ + namespace inpainting + { + struct tv + { + typedef float float_type; + }; + + template <typename T> + inline + void interval(T& v, + tv::float_type min, + tv::float_type max) + { + for (unsigned i = 0; i < mln_dim(T); ++i) + v[i] = std::max(std::min(v[i], max), min); + } + + template <typename I, typename M> + void tv(I& u, + const I& u_x, + const I& u_y, + const M& mask, + tv::float_type& max_diff_norm) + { + typedef tv::float_type float_type; + + extension::adjust_duplicate(u_x, 1); + extension::adjust_duplicate(u_y, 1); + + const float_type* dx_kern = make_dx_kernel<float_type>(); + const float_type* dy_kern = make_dy_kernel<float_type>(); + + static window2d full3x3; + static const bool vals[] = { 1, 1, 1, + 1, 1, 1, + 1, 1, 1 }; + + convert::from_to(vals, full3x3); + + mln_pixter(I) p(u); + mln_pixter(const M) p_m(mask); + mln_pixter(const I) p_x(u_x); + mln_pixter(const I) p_y(u_y); + + mln_qixter(const I, const window2d) q_x(p_x, full3x3); + mln_qixter(const I, const window2d) q_y(p_y, full3x3); + + for (p.start(), p_m.start(), p_x.start(), p_y.start(); + p.is_valid(); + p.next(), p_m.next(), p_x.next(), p_y.next()) + { + if (p_m.val()) + { + mln_value(I) old = p.val(); + + mln_value(I) p_x = literal::zero; + mln_value(I) p_y = literal::zero; + + unsigned k = 0; + for_all_2 (q_x, q_y) + { + p_x += q_x.val() * dx_kern[k]; + p_y += q_y.val() * dy_kern[k]; + ++k; + } + + mln_value(I) dv = p_x + p_y; + + p.val() += dv; + + interval(p.val(), 0.0, 255.0); + + float_type diff = norm::l2(old - p.val()); + + if (diff > max_diff_norm) + { + max_diff_norm = diff; + } + } + } + } + + } /* mln::inpainting */ + +} /* mln */ -- 1.7.2.5

1 0