* mln/fun/x2v/bilinear.hh: Add a new specialization for 3d Images. --- milena/ChangeLog | 6 ++++ milena/mln/fun/x2v/bilinear.hh | 65 ++++++++++++++++++++++++++++++++++++---- 2 files changed, 65 insertions(+), 6 deletions(-) diff --git a/milena/ChangeLog b/milena/ChangeLog index 93b5737..1c446cd 100644 --- a/milena/ChangeLog +++ b/milena/ChangeLog @@ -1,5 +1,11 @@ 2009-02-18 Guillaume Lazzara <z@lrde.epita.fr> + Add support for 3D images in fun::x2v::bilinear. + + * mln/fun/x2v/bilinear.hh: Add a new specialization for 3d Images. + +2009-02-18 Guillaume Lazzara <z@lrde.epita.fr> + Fix fun::x2v::rotation. * mln/fun/x2x/rotation.hh: diff --git a/milena/mln/fun/x2v/bilinear.hh b/milena/mln/fun/x2v/bilinear.hh index b0add18..e264462 100644 --- a/milena/mln/fun/x2v/bilinear.hh +++ b/milena/mln/fun/x2v/bilinear.hh @@ -1,4 +1,5 @@ -// Copyright (C) 2008 EPITA Research and Development Laboratory (LRDE) +// Copyright (C) 2008, 2009 EPITA Research and Development Laboratory +// (LRDE) // // This file is part of the Olena Library. This library is free // software; you can redistribute it and/or modify it under the terms @@ -59,9 +60,15 @@ namespace mln bilinear(const I& ima); - template <unsigned n, typename T> + /// Bilinear filtering on 2d images. + template <typename T> mln_value(I) - operator()(const algebra::vec<n,T>& v) const; + operator()(const algebra::vec<2,T>& v) const; + + /// Bilinear filtering on 3d images. Work on slices. + template <typename T> + mln_value(I) + operator()(const algebra::vec<3,T>& v) const; const I& ima; }; @@ -72,13 +79,13 @@ namespace mln template <typename I> bilinear<I>::bilinear(const I& ima) : ima(ima) { - mlc_bool(I::psite::dim == 2)::check(); + //mlc_bool(I::psite::dim == 2)::check(); } template <typename I> - template <unsigned n, typename T> + template <typename T> mln_value(I) - bilinear<I>::operator()(const algebra::vec<n,T>& v) const + bilinear<I>::operator()(const algebra::vec<2,T>& v) const { typedef mln_sum(mln_value(I)) vsum; @@ -121,6 +128,52 @@ namespace mln } + template <typename I> + template <typename T> + mln_value(I) + bilinear<I>::operator()(const algebra::vec<3,T>& v) const + { + typedef mln_sum(mln_value(I)) vsum; + + // q12----r2----q22 + // | | | + // | x | + // | | | + // q11----r1----q21 + + double x = v[1]; + double y = v[2]; + + double x1 = std::floor(x); + double x2 = std::floor(x) + 1; + double y1 = std::floor(y); + double y2 = std::floor(y) + 1; + def::coord z = math::round<float>()(v[0]); + + //Following access are supposed valid. + vsum q11 = ima(point3d(z, static_cast<unsigned>(x1), static_cast<unsigned>(y1))); + vsum q12 = ima(point3d(z, static_cast<unsigned>(x1), static_cast<unsigned>(y2))); + vsum q21 = ima(point3d(z, static_cast<unsigned>(x2), static_cast<unsigned>(y1))); + vsum q22 = ima(point3d(z, static_cast<unsigned>(x2), static_cast<unsigned>(y2))); + + double x2_x1 = x2 - x1; + double y2_y1 = y2 - y1; + + // linear interpolation #1 + vsum img_r1 = q11 * (x2 - x) / (x2_x1) + + q21 * (x - x1) / (x2_x1); + + // linear interpolation #2 + vsum img_r2 = q12 * (x2 - x) / (x2_x1) + q22 * (x - x1) / (x2_x1); + + // interpolating in y direction + vsum res = (img_r1 * (y2 - y) / (y2_y1) + + img_r2 * (y - y1) / (y2_y1)); + + return convert::to<mln_value(I)>(res); + } + + # endif // ! MLN_INCLUDE_ONLY } // end of namespace mln::fun::x2v -- 1.5.6.5