generic-library/opencv
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

removed ERFilter (to be moved to opencv_contrib/modules/text) and lineMOD (to be moved to opencv_contrib/modules/rgbd)

Browse Source
This commit is contained in:
Vadim Pisarevsky
2014-07-28 16:48:53 +04:00
parent 00263dd664
commit 1ad9827fc4
22 changed files with 0 additions and 7468 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
modules/objdetect/include/opencv2/objdetect.hpp
View File

@@ -315,8 +315,6 @@ public:
}
#include "opencv2/objdetect/linemod.hpp"
#include "opencv2/objdetect/erfilter.hpp"
#include "opencv2/objdetect/detection_based_tracker.hpp"
#endif

266
modules/objdetect/include/opencv2/objdetect/erfilter.hpp
View File

@@ -1,266 +0,0 @@
/*M///////////////////////////////////////////////////////////////////////////////////////
//
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
//
// By downloading, copying, installing or using the software you agree to this license.
// If you do not agree to this license, do not download, install,
// copy or use the software.
//
//
// License Agreement
// For Open Source Computer Vision Library
//
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
// Third party copyrights are property of their respective owners.
//
// Redistribution and use in source and binary forms, with or without modification,
// are permitted provided that the following conditions are met:
//
// * Redistribution's of source code must retain the above copyright notice,
// this list of conditions and the following disclaimer.
//
// * Redistribution's in binary form must reproduce the above copyright notice,
// this list of conditions and the following disclaimer in the documentation
// and/or other materials provided with the distribution.
//
// * The name of the copyright holders may not be used to endorse or promote products
// derived from this software without specific prior written permission.
//
// This software is provided by the copyright holders and contributors "as is" and
// any express or implied warranties, including, but not limited to, the implied
// warranties of merchantability and fitness for a particular purpose are disclaimed.
// In no event shall the Intel Corporation or contributors be liable for any direct,
// indirect, incidental, special, exemplary, or consequential damages
// (including, but not limited to, procurement of substitute goods or services;
// loss of use, data, or profits; or business interruption) however caused
// and on any theory of liability, whether in contract, strict liability,
// or tort (including negligence or otherwise) arising in any way out of
// the use of this software, even if advised of the possibility of such damage.
//
//M*/
#ifndef __OPENCV_OBJDETECT_ERFILTER_HPP__
#define __OPENCV_OBJDETECT_ERFILTER_HPP__
#include "opencv2/core.hpp"
#include <vector>
#include <deque>
#include <string>
namespace cv
{
/*!
Extremal Region Stat structure
The ERStat structure represents a class-specific Extremal Region (ER).
An ER is a 4-connected set of pixels with all its grey-level values smaller than the values
in its outer boundary. A class-specific ER is selected (using a classifier) from all the ER's
in the component tree of the image.
*/
struct CV_EXPORTS ERStat
{
public:
//! Constructor
explicit ERStat(int level = 256, int pixel = 0, int x = 0, int y = 0);
//! Destructor
~ERStat() { }
//! seed point and the threshold (max grey-level value)
int pixel;
int level;
//! incrementally computable features
int area;
int perimeter;
int euler; //!< euler number
Rect rect;
double raw_moments[2]; //!< order 1 raw moments to derive the centroid
double central_moments[3]; //!< order 2 central moments to construct the covariance matrix
std::deque<int> *crossings;//!< horizontal crossings
float med_crossings; //!< median of the crossings at three different height levels
//! 2nd stage features
float hole_area_ratio;
float convex_hull_ratio;
float num_inflexion_points;
// TODO Other features can be added (average color, standard deviation, and such)
// TODO shall we include the pixel list whenever available (i.e. after 2nd stage) ?
std::vector<int> *pixels;
//! probability that the ER belongs to the class we are looking for
double probability;
//! pointers preserving the tree structure of the component tree
ERStat* parent;
ERStat* child;
ERStat* next;
ERStat* prev;
//! wenever the regions is a local maxima of the probability
bool local_maxima;
ERStat* max_probability_ancestor;
ERStat* min_probability_ancestor;
};
/*!
Base class for 1st and 2nd stages of Neumann and Matas scene text detection algorithms
Neumann L., Matas J.: Real-Time Scene Text Localization and Recognition, CVPR 2012
Extracts the component tree (if needed) and filter the extremal regions (ER's) by using a given classifier.
*/
class CV_EXPORTS ERFilter : public Algorithm
{
public:
//! callback with the classifier is made a class. By doing it we hide SVM, Boost etc.
class CV_EXPORTS Callback
{
public:
virtual ~Callback() { }
//! The classifier must return probability measure for the region.
virtual double eval(const ERStat& stat) = 0; //const = 0; //TODO why cannot use const = 0 here?
};
/*!
the key method. Takes image on input and returns the selected regions in a vector of ERStat
only distinctive ERs which correspond to characters are selected by a sequential classifier
\param image is the input image
\param regions is output for the first stage, input/output for the second one.
*/
virtual void run( InputArray image, std::vector<ERStat>& regions ) = 0;
//! set/get methods to set the algorithm properties,
virtual void setCallback(const Ptr<ERFilter::Callback>& cb) = 0;
virtual void setThresholdDelta(int thresholdDelta) = 0;
virtual void setMinArea(float minArea) = 0;
virtual void setMaxArea(float maxArea) = 0;
virtual void setMinProbability(float minProbability) = 0;
virtual void setMinProbabilityDiff(float minProbabilityDiff) = 0;
virtual void setNonMaxSuppression(bool nonMaxSuppression) = 0;
virtual int getNumRejected() = 0;
};
/*!
Create an Extremal Region Filter for the 1st stage classifier of N&M algorithm
Neumann L., Matas J.: Real-Time Scene Text Localization and Recognition, CVPR 2012
The component tree of the image is extracted by a threshold increased step by step
from 0 to 255, incrementally computable descriptors (aspect_ratio, compactness,
number of holes, and number of horizontal crossings) are computed for each ER
and used as features for a classifier which estimates the class-conditional
probability P(er|character). The value of P(er|character) is tracked using the inclusion
relation of ER across all thresholds and only the ERs which correspond to local maximum
of the probability P(er|character) are selected (if the local maximum of the
probability is above a global limit pmin and the difference between local maximum and
local minimum is greater than minProbabilityDiff).
\param cb Callback with the classifier.
default classifier can be implicitly load with function loadClassifierNM1()
from file in samples/cpp/trained_classifierNM1.xml
\param thresholdDelta Threshold step in subsequent thresholds when extracting the component tree
\param minArea The minimum area (% of image size) allowed for retreived ER's
\param minArea The maximum area (% of image size) allowed for retreived ER's
\param minProbability The minimum probability P(er|character) allowed for retreived ER's
\param nonMaxSuppression Whenever non-maximum suppression is done over the branch probabilities
\param minProbability The minimum probability difference between local maxima and local minima ERs
*/
CV_EXPORTS Ptr<ERFilter> createERFilterNM1(const Ptr<ERFilter::Callback>& cb,
int thresholdDelta = 1, float minArea = 0.00025,
float maxArea = 0.13, float minProbability = 0.4,
bool nonMaxSuppression = true,
float minProbabilityDiff = 0.1);
/*!
Create an Extremal Region Filter for the 2nd stage classifier of N&M algorithm
Neumann L., Matas J.: Real-Time Scene Text Localization and Recognition, CVPR 2012
In the second stage, the ERs that passed the first stage are classified into character
and non-character classes using more informative but also more computationally expensive
features. The classifier uses all the features calculated in the first stage and the following
additional features: hole area ratio, convex hull ratio, and number of outer inflexion points.
\param cb Callback with the classifier
default classifier can be implicitly load with function loadClassifierNM2()
from file in samples/cpp/trained_classifierNM2.xml
\param minProbability The minimum probability P(er|character) allowed for retreived ER's
*/
CV_EXPORTS Ptr<ERFilter> createERFilterNM2(const Ptr<ERFilter::Callback>& cb,
float minProbability = 0.3);
/*!
Allow to implicitly load the default classifier when creating an ERFilter object.
The function takes as parameter the XML or YAML file with the classifier model
(e.g. trained_classifierNM1.xml) returns a pointer to ERFilter::Callback.
*/
CV_EXPORTS Ptr<ERFilter::Callback> loadClassifierNM1(const std::string& filename);
/*!
Allow to implicitly load the default classifier when creating an ERFilter object.
The function takes as parameter the XML or YAML file with the classifier model
(e.g. trained_classifierNM1.xml) returns a pointer to ERFilter::Callback.
*/
CV_EXPORTS Ptr<ERFilter::Callback> loadClassifierNM2(const std::string& filename);
// computeNMChannels operation modes
enum { ERFILTER_NM_RGBLGrad = 0,
ERFILTER_NM_IHSGrad = 1
};
/*!
Compute the different channels to be processed independently in the N&M algorithm
Neumann L., Matas J.: Real-Time Scene Text Localization and Recognition, CVPR 2012
In N&M algorithm, the combination of intensity (I), hue (H), saturation (S), and gradient
magnitude channels (Grad) are used in order to obtain high localization recall.
This implementation also provides an alternative combination of red (R), green (G), blue (B),
lightness (L), and gradient magnitude (Grad).
\param _src Source image. Must be RGB CV_8UC3.
\param _channels Output vector<Mat> where computed channels are stored.
\param _mode Mode of operation. Currently the only available options are
ERFILTER_NM_RGBLGrad (by default) and ERFILTER_NM_IHSGrad.
*/
CV_EXPORTS void computeNMChannels(InputArray _src, OutputArrayOfArrays _channels, int _mode = ERFILTER_NM_RGBLGrad);
/*!
Find groups of Extremal Regions that are organized as text blocks. This function implements
the grouping algorithm described in:
Gomez L. and Karatzas D.: Multi-script Text Extraction from Natural Scenes, ICDAR 2013.
Notice that this implementation constrains the results to horizontally-aligned text and
latin script (since ERFilter classifiers are trained only for latin script detection).
The algorithm combines two different clustering techniques in a single parameter-free procedure
to detect groups of regions organized as text. The maximally meaningful groups are fist detected
in several feature spaces, where each feature space is a combination of proximity information
(x,y coordinates) and a similarity measure (intensity, color, size, gradient magnitude, etc.),
thus providing a set of hypotheses of text groups. Evidence Accumulation framework is used to
combine all these hypotheses to get the final estimate. Each of the resulting groups are finally
validated using a classifier in order to assest if they form a valid horizontally-aligned text block.
\param src Vector of sinle channel images CV_8UC1 from wich the regions were extracted.
\param regions Vector of ER's retreived from the ERFilter algorithm from each channel
\param filename The XML or YAML file with the classifier model (e.g. trained_classifier_erGrouping.xml)
\param minProbability The minimum probability for accepting a group
\param groups The output of the algorithm are stored in this parameter as list of rectangles.
*/
CV_EXPORTS void erGrouping(InputArrayOfArrays src, std::vector<std::vector<ERStat> > &regions,
const std::string& filename, float minProbablity,
std::vector<Rect > &groups);
}
#endif // _OPENCV_ERFILTER_HPP_

455
modules/objdetect/include/opencv2/objdetect/linemod.hpp
View File

@@ -1,455 +0,0 @@
/*M///////////////////////////////////////////////////////////////////////////////////////
//
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
//
// By downloading, copying, installing or using the software you agree to this license.
// If you do not agree to this license, do not download, install,
// copy or use the software.
//
//
// License Agreement
// For Open Source Computer Vision Library
//
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
// Third party copyrights are property of their respective owners.
//
// Redistribution and use in source and binary forms, with or without modification,
// are permitted provided that the following conditions are met:
//
// * Redistribution's of source code must retain the above copyright notice,
// this list of conditions and the following disclaimer.
//
// * Redistribution's in binary form must reproduce the above copyright notice,
// this list of conditions and the following disclaimer in the documentation
// and/or other materials provided with the distribution.
//
// * The name of the copyright holders may not be used to endorse or promote products
// derived from this software without specific prior written permission.
//
// This software is provided by the copyright holders and contributors "as is" and
// any express or implied warranties, including, but not limited to, the implied
// warranties of merchantability and fitness for a particular purpose are disclaimed.
// In no event shall the Intel Corporation or contributors be liable for any direct,
// indirect, incidental, special, exemplary, or consequential damages
// (including, but not limited to, procurement of substitute goods or services;
// loss of use, data, or profits; or business interruption) however caused
// and on any theory of liability, whether in contract, strict liability,
// or tort (including negligence or otherwise) arising in any way out of
// the use of this software, even if advised of the possibility of such damage.
//
//M*/
#ifndef __OPENCV_OBJDETECT_LINEMOD_HPP__
#define __OPENCV_OBJDETECT_LINEMOD_HPP__
#include "opencv2/core.hpp"
#include <map>
/****************************************************************************************\
* LINE-MOD *
\****************************************************************************************/
namespace cv {
namespace linemod {
/// @todo Convert doxy comments to rst
/**
* \brief Discriminant feature described by its location and label.
*/
struct CV_EXPORTS Feature
{
int x; ///< x offset
int y; ///< y offset
int label; ///< Quantization
Feature() : x(0), y(0), label(0) {}
Feature(int x, int y, int label);
void read(const FileNode& fn);
void write(FileStorage& fs) const;
};
inline Feature::Feature(int _x, int _y, int _label) : x(_x), y(_y), label(_label) {}
struct CV_EXPORTS Template
{
int width;
int height;
int pyramid_level;
std::vector<Feature> features;
void read(const FileNode& fn);
void write(FileStorage& fs) const;
};
/**
* \brief Represents a modality operating over an image pyramid.
*/
class QuantizedPyramid
{
public:
// Virtual destructor
virtual ~QuantizedPyramid() {}
/**
* \brief Compute quantized image at current pyramid level for online detection.
*
* \param[out] dst The destination 8-bit image. For each pixel at most one bit is set,
* representing its classification.
*/
virtual void quantize(Mat& dst) const =0;
/**
* \brief Extract most discriminant features at current pyramid level to form a new template.
*
* \param[out] templ The new template.
*/
virtual bool extractTemplate(Template& templ) const =0;
/**
* \brief Go to the next pyramid level.
*
* \todo Allow pyramid scale factor other than 2
*/
virtual void pyrDown() =0;
protected:
/// Candidate feature with a score
struct Candidate
{
Candidate(int x, int y, int label, float score);
/// Sort candidates with high score to the front
bool operator<(const Candidate& rhs) const
{
return score > rhs.score;
}
Feature f;
float score;
};
/**
* \brief Choose candidate features so that they are not bunched together.
*
* \param[in] candidates Candidate features sorted by score.
* \param[out] features Destination vector of selected features.
* \param[in] num_features Number of candidates to select.
* \param[in] distance Hint for desired distance between features.
*/
static void selectScatteredFeatures(const std::vector<Candidate>& candidates,
std::vector<Feature>& features,
size_t num_features, float distance);
};
inline QuantizedPyramid::Candidate::Candidate(int x, int y, int label, float _score) : f(x, y, label), score(_score) {}
/**
* \brief Interface for modalities that plug into the LINE template matching representation.
*
* \todo Max response, to allow optimization of summing (255/MAX) features as uint8
*/
class CV_EXPORTS Modality
{
public:
// Virtual destructor
virtual ~Modality() {}
/**
* \brief Form a quantized image pyramid from a source image.
*
* \param[in] src The source image. Type depends on the modality.
* \param[in] mask Optional mask. If not empty, unmasked pixels are set to zero
* in quantized image and cannot be extracted as features.
*/
Ptr<QuantizedPyramid> process(const Mat& src,
const Mat& mask = Mat()) const
{
return processImpl(src, mask);
}
virtual String name() const =0;
virtual void read(const FileNode& fn) =0;
virtual void write(FileStorage& fs) const =0;
/**
* \brief Create modality by name.
*
* The following modality types are supported:
* - "ColorGradient"
* - "DepthNormal"
*/
static Ptr<Modality> create(const String& modality_type);
/**
* \brief Load a modality from file.
*/
static Ptr<Modality> create(const FileNode& fn);
protected:
// Indirection is because process() has a default parameter.
virtual Ptr<QuantizedPyramid> processImpl(const Mat& src,
const Mat& mask) const =0;
};
/**
* \brief Modality that computes quantized gradient orientations from a color image.
*/
class CV_EXPORTS ColorGradient : public Modality
{
public:
/**
* \brief Default constructor. Uses reasonable default parameter values.
*/
ColorGradient();
/**
* \brief Constructor.
*
* \param weak_threshold When quantizing, discard gradients with magnitude less than this.
* \param num_features How many features a template must contain.
* \param strong_threshold Consider as candidate features only gradients whose norms are
* larger than this.
*/
ColorGradient(float weak_threshold, size_t num_features, float strong_threshold);
virtual String name() const;
virtual void read(const FileNode& fn);
virtual void write(FileStorage& fs) const;
float weak_threshold;
size_t num_features;
float strong_threshold;
protected:
virtual Ptr<QuantizedPyramid> processImpl(const Mat& src,
const Mat& mask) const;
};
/**
* \brief Modality that computes quantized surface normals from a dense depth map.
*/
class CV_EXPORTS DepthNormal : public Modality
{
public:
/**
* \brief Default constructor. Uses reasonable default parameter values.
*/
DepthNormal();
/**
* \brief Constructor.
*
* \param distance_threshold Ignore pixels beyond this distance.
* \param difference_threshold When computing normals, ignore contributions of pixels whose
* depth difference with the central pixel is above this threshold.
* \param num_features How many features a template must contain.
* \param extract_threshold Consider as candidate feature only if there are no differing
* orientations within a distance of extract_threshold.
*/
DepthNormal(int distance_threshold, int difference_threshold, size_t num_features,
int extract_threshold);
virtual String name() const;
virtual void read(const FileNode& fn);
virtual void write(FileStorage& fs) const;
int distance_threshold;
int difference_threshold;
size_t num_features;
int extract_threshold;
protected:
virtual Ptr<QuantizedPyramid> processImpl(const Mat& src,
const Mat& mask) const;
};
/**
* \brief Debug function to colormap a quantized image for viewing.
*/
void colormap(const Mat& quantized, Mat& dst);
/**
* \brief Represents a successful template match.
*/
struct CV_EXPORTS Match
{
Match()
{
}
Match(int x, int y, float similarity, const String& class_id, int template_id);
/// Sort matches with high similarity to the front
bool operator<(const Match& rhs) const
{
// Secondarily sort on template_id for the sake of duplicate removal
if (similarity != rhs.similarity)
return similarity > rhs.similarity;
else
return template_id < rhs.template_id;
}
bool operator==(const Match& rhs) const
{
return x == rhs.x && y == rhs.y && similarity == rhs.similarity && class_id == rhs.class_id;
}
int x;
int y;
float similarity;
String class_id;
int template_id;
};
inline
Match::Match(int _x, int _y, float _similarity, const String& _class_id, int _template_id)
: x(_x), y(_y), similarity(_similarity), class_id(_class_id), template_id(_template_id)
{}
/**
* \brief Object detector using the LINE template matching algorithm with any set of
* modalities.
*/
class CV_EXPORTS Detector
{
public:
/**
* \brief Empty constructor, initialize with read().
*/
Detector();
/**
* \brief Constructor.
*
* \param modalities Modalities to use (color gradients, depth normals, ...).
* \param T_pyramid Value of the sampling step T at each pyramid level. The
* number of pyramid levels is T_pyramid.size().
*/
Detector(const std::vector< Ptr<Modality> >& modalities, const std::vector<int>& T_pyramid);
/**
* \brief Detect objects by template matching.
*
* Matches globally at the lowest pyramid level, then refines locally stepping up the pyramid.
*
* \param sources Source images, one for each modality.
* \param threshold Similarity threshold, a percentage between 0 and 100.
* \param[out] matches Template matches, sorted by similarity score.
* \param class_ids If non-empty, only search for the desired object classes.
* \param[out] quantized_images Optionally return vector<Mat> of quantized images.
* \param masks The masks for consideration during matching. The masks should be CV_8UC1
* where 255 represents a valid pixel. If non-empty, the vector must be
* the same size as sources. Each element must be
* empty or the same size as its corresponding source.
*/
void match(const std::vector<Mat>& sources, float threshold, std::vector<Match>& matches,
const std::vector<String>& class_ids = std::vector<String>(),
OutputArrayOfArrays quantized_images = noArray(),
const std::vector<Mat>& masks = std::vector<Mat>()) const;
/**
* \brief Add new object template.
*
* \param sources Source images, one for each modality.
* \param class_id Object class ID.
* \param object_mask Mask separating object from background.
* \param[out] bounding_box Optionally return bounding box of the extracted features.
*
* \return Template ID, or -1 if failed to extract a valid template.
*/
int addTemplate(const std::vector<Mat>& sources, const String& class_id,
const Mat& object_mask, Rect* bounding_box = NULL);
/**
* \brief Add a new object template computed by external means.
*/
int addSyntheticTemplate(const std::vector<Template>& templates, const String& class_id);
/**
* \brief Get the modalities used by this detector.
*
* You are not permitted to add/remove modalities, but you may dynamic_cast them to
* tweak parameters.
*/
const std::vector< Ptr<Modality> >& getModalities() const { return modalities; }
/**
* \brief Get sampling step T at pyramid_level.
*/
int getT(int pyramid_level) const { return T_at_level[pyramid_level]; }
/**
* \brief Get number of pyramid levels used by this detector.
*/
int pyramidLevels() const { return pyramid_levels; }
/**
* \brief Get the template pyramid identified by template_id.
*
* For example, with 2 modalities (Gradient, Normal) and two pyramid levels
* (L0, L1), the order is (GradientL0, NormalL0, GradientL1, NormalL1).
*/
const std::vector<Template>& getTemplates(const String& class_id, int template_id) const;
int numTemplates() const;
int numTemplates(const String& class_id) const;
int numClasses() const { return static_cast<int>(class_templates.size()); }
std::vector<String> classIds() const;
void read(const FileNode& fn);
void write(FileStorage& fs) const;
String readClass(const FileNode& fn, const String &class_id_override = "");
void writeClass(const String& class_id, FileStorage& fs) const;
void readClasses(const std::vector<String>& class_ids,
const String& format = "templates_%s.yml.gz");
void writeClasses(const String& format = "templates_%s.yml.gz") const;
protected:
std::vector< Ptr<Modality> > modalities;
int pyramid_levels;
std::vector<int> T_at_level;
typedef std::vector<Template> TemplatePyramid;
typedef std::map<String, std::vector<TemplatePyramid> > TemplatesMap;
TemplatesMap class_templates;
typedef std::vector<Mat> LinearMemories;
// Indexed as [pyramid level][modality][quantized label]
typedef std::vector< std::vector<LinearMemories> > LinearMemoryPyramid;
void matchClass(const LinearMemoryPyramid& lm_pyramid,
const std::vector<Size>& sizes,
float threshold, std::vector<Match>& matches,
const String& class_id,
const std::vector<TemplatePyramid>& template_pyramids) const;
};
/**
* \brief Factory function for detector using LINE algorithm with color gradients.
*
* Default parameter settings suitable for VGA images.
*/
CV_EXPORTS Ptr<Detector> getDefaultLINE();
/**
* \brief Factory function for detector using LINE-MOD algorithm with color gradients
* and depth normals.
*
* Default parameter settings suitable for VGA images.
*/
CV_EXPORTS Ptr<Detector> getDefaultLINEMOD();
} // namespace linemod
} // namespace cv
#endif // __OPENCV_OBJDETECT_LINEMOD_HPP__