Olena March 2004

olena@lrde.epita.fr

15 participants
44 discussions

[color_morpho] pour Theo, Giovanni et les curieux
by Niels Van Vliet 30 Mar '04

30 Mar '04

Nous venons de créer le projet prcs color_morpho. Ce dernier vise à rassembler le travail que je fais avec Gioavanni sur la segmentation d'images couleurs. Ce n'est pas propre (c'est plus un joli goinfre qu'un projet). Pour l'instant il n'y a évidemment pas grand chose. Pour Théo et Giovanni: 0) [prcs checkout oln] 1) prcs checkout color_morpho 2) ./configure.sh 3) make 4) // pause café 5) cd scripts 6) ./for_all_full_process.sh < lst_img.conf 7) // pause clope 8) gqview ../data/out [Heu, il me semble qu'il y a un bug dans l'attribut min_value, je regarde ça] -- Niels

1 0

Release Olena
by Damien Thivolle 25 Mar '04

25 Mar '04

Suite à la réunion d'aujourd'hui midi à laquelle étaient absents : Simon et Niels (c'est mal :p), les décisions suivantes ont été prises : Concernant la Doc : 2 versions, une daily-buildée, sera mise sur l'intra, et une pour la release, sera mise sur le wiki. La documentation de integre et metalic sera faite pour la prochaine release. Concernant la release : La date du 15 avril a été retenue et j'ai été désigné pour etre le release manager. Il faudrait qu'olena compile sous mac, debian, *bsd, cygwin, voire sun et windows, et avec les compilateurs suivants : g++3.2, g++3.3 icc7, icc8, como, visual c++ .net. Il faudrait que tout le monde mette à jour ses assignments sur le wiki et que chacun donne ses idées sur le contenu de la release suivante. -- Damien Thivolle damien.thivolle(a)lrde.epita.fr

3 2

MPOOL'04
by Thierry Géraud 25 Mar '04

25 Mar '04

lu sur http://www.ifi.uio.no/ecoop2004/ws-index.html : The following workshops are part of the 18th European Conference on Object-Oriented Programming (ECOOP 2004) in Oslo. WS11: Multiparadigm Programming with Object-Oriented Languages (MPOOL'04) WS12: Mechanisms for Specialization, Generalization and Inheritance (MASPEGHI 2004) WS18: Ninth International Workshop on Component-Oriented Programming (WCOP 2004) WS21: ECOOP Workshop on Programming Languages and Operating Systems (ECOOP-PLOS 2004) WS22: 5th International Workshop on Object-Oriented Reengineering (OOR 2004)

1 0

Idees Prochaine Realease Olena
by DARBON Jerome 25 Mar '04

25 Mar '04

"Geometric Active Contours" : dans le cadre dde ma these.

1 1

réunion demain jeudi 12h
by Thierry Géraud 24 Mar '04

24 Mar '04

ordres du jour : - planifier la prochaine release - établir des listes de tâches par priorité décroissante - mise à jour du wiki

1 0

réunion mercredi 9h
by Simon Odou 23 Mar '04

23 Mar '04

Tout est dans le titre... -- Simon Odou simon(a)lrde.epita.fr

1 0

Corel Draw and Corel Ventura 242999
by 242999＠bigfoot.com 21 Mar '04

21 Mar '04

1 0

Discount code 242992
by 242992＠excite.com 21 Mar '04

21 Mar '04

1 0

oln 10.83: doc (utils)
by Niels Van Vliet 19 Mar '04

19 Mar '04

Index: olena/oln/utils/copy.hh --- olena/oln/utils/copy.hh Mon, 25 Aug 2003 11:47:33 +0200 burrus_n (oln/i/29_copy.hh 1.7 640) +++ olena/oln/utils/copy.hh Sat, 13 Mar 2004 18:06:17 +0100 van-vl_n (oln/i/29_copy.hh 1.7 640) @@ -1,4 +1,4 @@ -// Copyright (C) 2001, 2002, 2003 EPITA Research and Development Laboratory +// Copyright (C) 2001, 2002, 2003, 2004 EPITA Research and Development Laboratory // // This file is part of the Olena Library. This library is free // software; you can redistribute it and/or modify it under the terms @@ -28,15 +28,18 @@ #ifndef OLENA_UTILS_COPY_HH # define OLENA_UTILS_COPY_HH -// This file contains implementations of image's copy constructors -// -// / \ NEVER include this file unless you understand what -// / ! \ it does and why you need it. -// `-----' -// -// FIXME: should be replaced soon by a correct const image handling. -// a detailed problem description can be found on: -// http://www.lrde.epita.fr/cgi-bin/twiki/view/Projects/OlenaConstImages +/* \file copy.hh +** +** \warning This file contains implementations of image's copy constructors. +**\verbatim +** / \ NEVER include this file unless you understand what +** / ! \ it does and why you need it. +** `-----' +**\endverbatim +** \todo FIXME: should be replaced soon by a correct const image handling. +** a detailed problem description can be found on: +** http://www.lrde.epita.fr/cgi-bin/twiki/view/Projects/OlenaConstImages +*/ namespace oln { Index: olena/oln/utils/histogram.hh --- olena/oln/utils/histogram.hh Wed, 11 Feb 2004 11:51:35 +0100 van-vl_n (oln/10_histogram. 1.6.1.14.1.6 640) +++ olena/oln/utils/histogram.hh Sat, 13 Mar 2004 21:17:34 +0100 van-vl_n (oln/10_histogram. 1.6.1.14.1.6 640) @@ -39,11 +39,12 @@ # include <vector> namespace oln { - + /// Namespace for utilities, such as statistic. namespace utils { + /// Function within this namespace should not be used outside this file. namespace internal { - //! Return the size of the space needed to explore the type T + //! Return the size of the space needed to explore the type T. template <typename T> struct img_max_size { @@ -82,11 +83,13 @@ }; } // end of namespace internal + /// Namespace for abstract utilities. namespace abstract { - // FIXME: An image is inside the histogram. This is incorrect - // because it is not exactly an image (we do not need any - // border). + /*! Abstract base class for historgram. + ** + ** \see oln::histogram + */ template<class T, typename CPT, class Exact = mlc::final> @@ -96,25 +99,28 @@ typedef histogram<T, CPT, Exact> self_type; typedef Exact exact_type; typedef CPT cpt_type; - + /// Put the number of occurrence of every value to zero. void clear() { this->exact().clear_impl(); } - + /// Read the number of occurrence of \a v. const CPT& operator[](const T &v)const { return this->exact().at(v); } - + /// Read or write the number of occurrence of \a v. CPT& operator[](const T &v) { return this->exact().at(v); } - + /*! Build the histogram of an image. + ** + ** \attention The histogram is not cleared. + */ template <class I> void init(const oln::abstract::image &img) @@ -124,6 +130,43 @@ }; } // end of namespace abstract + /*! Histogram. + ** + ** This histogram uses an image of unsigned to store the value. + ** For example the histogram of an image<int_u8> will store the + ** number of occurrences in an image1d; an image<rgb_8> will store the + ** number of occurrences an image3d (because rgb_8 has 3 components). + ** + ** \note FIXME: An image is inside the histogram. This is incorrect + ** because it is not exactly an image (we do not need any border). + ** + ** \param T Type of the image. + ** \param CPT Type use to count the occurrences (unsinged). + ** \param V2P Concersion class to convert a value T to a point. + ** \param Exact Exact type of the histogram. + ** + ** \see oln::abstract::histogram + ** + ** \code + ** #include <oln/basics2d.hh> + ** #include <oln/utils/histogram.hh> + ** #include <ntg/all.hh> + ** #include <iostream> + ** + ** int main() + ** { + ** oln::image2d<ntg::rgb_8> in = oln::io::load(IMG_IN "lena.ppm"); + ** + ** oln::utils::histogram<ntg::rgb_8> h(in); + ** + ** ntg::rgb_8 pink(215, 129, 113); + ** + ** // h[pink] = 14 + ** std::cout << "Number of occurrences of the color (213, 129, 135): " + ** << h[pink] << std::endl; + ** } + ** \endcode + */ template<typename T, typename CPT = unsigned, class V2P = oln::convert::value_to_point<T>, @@ -142,12 +185,14 @@ typedef abstract::histogram<T, CPT, self_type> upper_type; typedef typename dim_traits<dim, CPT>::img_type img_type; + /// Empty histogram. The function Init(image) should be used after this constructor. histogram(const value_to_point_type & c2p = value_to_point_type()): v2p_(c2p), img_(internal::img_max_size<input_type>()()) { clear(); } + /// This compute the histogram of an image. template <class I> histogram(const oln::abstract::image & input, const value_to_point_type & v2p = value_to_point_type()): @@ -157,6 +202,7 @@ init(input); } + /// clear() should be called. void clear_impl() { @@ -165,18 +211,19 @@ img_[it] = ntg_zero_val(CPT); } + /// operator[] should be called. const CPT& at(const T &v)const { return img_[v2p_(v)]; } - + /// operator[] should be called. CPT& at(const T &v) { return img_[v2p_(v)]; } - + /// impl() should be called. template <class I> void init_impl(const oln::abstract::image &img) @@ -187,6 +234,7 @@ ++img_[v2p_(img[it_img])]; } + /// Return the image of occurrence. const img_type & image() const { @@ -198,7 +246,16 @@ img_type img_; }; - //Note: If there is no min an assertion will fail at the end of the loop. + /*! Minimum value of an histogram. + ** + ** It return the smaller value within the image used to build the histogram. + ** + ** \warning If there is no min an assertion will fail at the end of the loop. + ** + ** \note It can be slow when the histogram is sparse because it iterate over a + ** large range of 0. Use histogram_min or histogram_minmax instead. + ** \see histogram_min + */ template< typename T, typename CPT, class Exact> inline T min(const abstract::histogram<T, CPT, Exact>& hist) @@ -211,7 +268,16 @@ break; return i; } - //Note: If there is no max an assertion will fail at the end of the loop. + /*! Maximum value of an histogram. + ** + ** It return the higher value within the image used to build the histogram. + ** + ** \warning If there is no min an assertion will fail at the end of the loop. + ** + ** \note It can be slow when the histogram is sparse because it iterate over a + ** large range of 0. Use histogram_max or histogram_minmax instead. + ** \see histogram_max + */ template< typename T, typename CPT, class Exact> inline T max(const abstract::histogram<T, CPT, Exact>& hist) @@ -225,26 +291,21 @@ return i; } -/* The two functions above can be slow when the histogram is sparse - because they iterate over a large range of 0. - - It's inefficient to call them repeatedly while updating the - histogram (that's typically the case in the morpho::fast:: - algorithms). Better use the specialized histograms below. - - --- - - The idea behind the min- and max-specialized histogram is to - maintain worst min and max bounds while updating histogram. We - don't maintain _exact_ min and max bounds, because this would - involve some costly computation when removing values from the - histogram and maybe this time will be lost if the removed value is - reinsterted before max() or min() is called. - - So we update the _worst_ min and max bounds whenever the histogram - value are accessed, and delay the _real_ min and max compuation - until min() or max() is called. */ - + /*! Build the histogram and has quick min and max functions. + ** + ** The idea behind the min- and max-specialized histogram is to + ** maintain worst min and max bounds while updating histogram. We + ** don't maintain _exact_ min and max bounds, because this would + ** involve some costly computation when removing values from the + ** histogram and maybe this time will be lost if the removed value is + ** reinserted before max() or min() is called.\n + ** So we update the _worst_ min and max bounds whenever the histogram + ** value are accessed, and delay the _real_ min and max computation + ** until min() or max() is called. + ** \see histogram + ** \see histogram_min + ** \see histogram_max + */ template< typename T, typename CPT = unsigned, class V2P = convert::value_to_point<T>, @@ -256,6 +317,7 @@ private: typedef typename ntg_is_a(T, ntg::non_vectorial)::ensure_type ensure_type; protected: + /// Maintain the worst min and max. void adjust(const T &idx) { @@ -280,7 +342,7 @@ upper_type(input, v2p), min_(ntg_min_val(T)), max_(ntg_max_val(T)) {} - + /// operator[] should be called. const CPT& at(const T& i) const { @@ -288,6 +350,7 @@ return img_[v2p_(i)]; } + /// operator[] should be called. CPT& at(const T& i) { @@ -295,6 +358,7 @@ return img_[v2p_(i)]; } + /// Quick function min. T min() { @@ -303,7 +367,7 @@ break; return min_; } - + /// Quick function max. T max() { @@ -314,9 +378,12 @@ } protected: - T min_, max_; // indices of min and max elements + T min_, max_; ///< Indices of min and max elements. }; - + /*! Build the histogram and has quick min function. + ** + ** \see histogram_minmax + */ template< typename T, typename CPT = unsigned, class V2P = convert::value_to_point<T>, @@ -328,6 +395,7 @@ private: typedef typename ntg_is_a(T, ntg::non_vectorial)::ensure_type ensure_type; protected: + /// Maintain the worst min. void adjust(const T& idx) { @@ -349,20 +417,21 @@ const value_to_point_type & v2p = value_to_point_type()) : upper_type(input, v2p), min_(ntg_min_val(T)) {} - + /// operator[] should be called. const CPT& at(const T& i) const { return img_[v2p_(i)]; } + /// operator[] should be called. CPT& at(const T& i) { adjust(i); return img_[v2p_(i)]; } - + /// Quick function min. T min() { @@ -371,7 +440,7 @@ break; return min_; } - + /// Return the min. T res() { @@ -379,9 +448,13 @@ } protected: - T min_; // index of min element + T min_; ///< Index of the worst min element. }; + /*! Build the histogram and has quick max function. + ** + ** \see histogram_minmax + */ template< typename T, typename CPT = unsigned, class V2P = convert::value_to_point<T> , @@ -391,6 +464,7 @@ Exact>::ret> { protected: + /// Maintain the worst max. void adjust(const T& idx) { @@ -412,19 +486,21 @@ const value_to_point_type & v2p = value_to_point_type()) : upper_type(input, v2p),max_(ntg_max_val(T)) {} + /// operator[] should be called. const CPT& at(const T& i) const { return img_[v2p_(i)]; } + /// operator[] should be called. CPT& at(const T& i) { adjust(i); return img_[v2p_(i)]; } - + /// Quick function max. T max() { @@ -433,7 +509,7 @@ break; return max_; } - + /// Return the max. T res() { @@ -441,34 +517,45 @@ } protected: - T max_; // index of max element + T max_; ///< Index of the worst max element. }; + /// Minimum non-zero value of an histogram. template< typename T, typename CPT, class V2P, class Exact > inline T min(histogram_minmax<T, CPT, V2P, Exact>& hist) { return hist.min(); } + /// Minimum non-zero value of an histogram. template< typename T, typename CPT, class V2P, class Exact > inline T min(histogram_min<T, CPT, V2P, Exact>& hist) { return hist.min(); } + /// Maximum non-zero value of an histogram. template< typename T, typename CPT, class V2P, class Exact > inline T max(histogram_minmax<T, CPT, V2P, Exact>& hist) { return hist.max(); } + /// Maximum non-zero value of an histogram. template< typename T, typename CPT, class V2P, class Exact > inline T max(histogram_max<T, CPT, V2P, Exact>& hist) { return hist.max(); } - + /*! Sort the values of an image, and store the result in a vector + ** + ** This sort is efficient. + ** \arg im Image non_vectorial. + ** \arg v sorted vector at the end of the function. + ** \pre precondition(v.size() == im.npoints()); + ** \pre A histogram of the image has to be possible. + */ template<class I> void distrib_sort(const oln::abstract::image& im, @@ -501,6 +588,14 @@ *(ptr[unsigned(im[p] - ntg_min_val(val))]++) = p; } + /*! Inverted sort of the values of an image, and store the result in a vector + ** + ** This sort is efficient. + ** \arg im Image non_vectorial. + ** \arg v sorted vector at the end of the function. + ** \pre precondition(v.size() == im.npoints()); + ** \pre A histogram of the image has to be possible. + */ template<class I> void distrib_sort_inv(const oln::abstract::image& im, @@ -526,7 +621,10 @@ *(ptr[unsigned(im[p] - ntg_min_val(val))]++) = p; } - // to select staticly the good distrib_sort + /*! Select staticly the good distrib_sort. + ** + ** \param reverse If the sort should be reverted or not. + */ template <bool reverse> struct select_distrib_sort { Index: olena/oln/utils/stat.hh --- olena/oln/utils/stat.hh Thu, 07 Aug 2003 02:37:23 +0200 burrus_n (oln/9_stat.hh 1.5.1.9 640) +++ olena/oln/utils/stat.hh Sat, 13 Mar 2004 20:50:57 +0100 van-vl_n (oln/9_stat.hh 1.5.1.9 640) @@ -1,4 +1,4 @@ -// Copyright (C) 2001, 2002, 2003 EPITA Research and Development Laboratory +// Copyright (C) 2001, 2002, 2003, 2004 EPITA Research and Development Laboratory // // This file is part of the Olena Library. This library is free // software; you can redistribute it and/or modify it under the terms @@ -35,6 +35,7 @@ namespace utils { + /// Unary function that stores the min and the max. template< class T > struct f_minmax : std::unary_function< const T&, void > { @@ -61,25 +62,28 @@ count_ = 0; } + /// True if a value has been tested. bool valued() const { return count_; } + /// Number of value has been tested. size_t count() const { return count_; } + /// Minimum found. const T min() const { assertion(valued()); return min_; } - + /// Maximum found. const T max() const { @@ -93,6 +97,7 @@ T max_; }; + /// Computes the mean, the variance and store the min, the max. template< class T, class C = ntg::float_s > struct f_moments : f_minmax<T> { @@ -114,13 +119,14 @@ super::operator()(val); } + /// Return the mean value. const C mean() const { assertion(this->valued()); return sum1_ / C(this->count()); } - + /// Return the variance. const C variance() const { Index: olena/oln/utils/timer.hh --- olena/oln/utils/timer.hh Fri, 07 Nov 2003 17:34:52 +0100 burrus_n (oln/8_timer.hh 1.8 640) +++ olena/oln/utils/timer.hh Sat, 13 Mar 2004 21:21:29 +0100 van-vl_n (oln/8_timer.hh 1.8 640) @@ -1,4 +1,4 @@ -// Copyright (C) 2001, 2002, 2003 EPITA Research and Development Laboratory +// Copyright (C) 2001, 2002, 2003, 2004 EPITA Research and Development Laboratory // // This file is part of the Olena Library. This library is free // software; you can redistribute it and/or modify it under the terms @@ -34,10 +34,16 @@ namespace oln { namespace utils { - + /*! Timer class. + */ class timer { public: + /*! Constructor. + ** + ** \note The timer is not stopped, it is in a so-called 'unknown state'. + ** Therefor, Restart can not be called. + */ timer() { status_ = e_unknown; @@ -45,6 +51,10 @@ start_time_ = clock(); } + /*! Start the timer. + ** + ** \pre The timer should not be running. + */ void start() { @@ -54,7 +64,10 @@ status_ = e_running; start_time_ = clock(); } - + /*! Restart the timer. + ** + ** \pre The timer should have been start at least one time. + */ float restart() { @@ -64,6 +77,10 @@ return val; } + /*! Resume the timer. + ** + ** \pre The timer should be running. + */ void resume() { @@ -72,6 +89,10 @@ start_time_ = clock(); } + /*! Stop the timer. + ** + ** \pre The timer should be running. + */ float stop() { @@ -82,7 +103,7 @@ return total_time(); } - // Time since the last resume() or start() + /// Time since the last resume() or start(). float last_time() const { @@ -93,6 +114,7 @@ float((clock() - start_time_)) / CLOCKS_PER_SEC; } + /// Total time elapsed. float total_time() const { Index: olena/oln/utils/special_points.hh --- olena/oln/utils/special_points.hh Tue, 10 Feb 2004 19:49:03 +0100 van-vl_n (oln/j/50_special_po 1.1 644) +++ olena/oln/utils/special_points.hh Sat, 13 Mar 2004 21:22:12 +0100 van-vl_n (oln/j/50_special_po 1.1 644) @@ -33,7 +33,7 @@ namespace oln { namespace utils { - + /// oln::utils::internal namespace internal { /*! Creates a point that is used as a state. *

2 5

slow or fast?
by Niels Van Vliet 18 Mar '04

18 Mar '04

In attribute_closing_opening_map.hh: ** im1 = oln::morpho::slow::card_closing(im1, oln::neighb_c4(), 200); ** oln::save(im1, IMG_OUT "oln_morpho_fast_card_closing.ppm"); Dans attribute_closing_opening_map.hh, ne faut-il pas changer les noms de fichiers *fast*.ppm en *slow*.ppm? -- Niels

2 1

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

Olena March 2004