OpenCV  4.1.1-pre
Open Source Computer Vision
Looking for a C++ dev who knows OpenCV?
I'm looking for work. Hire me!
cv::SparseMat Class Reference

The class SparseMat represents multi-dimensional sparse numerical arrays. More...

#include <opencv2/core/mat.hpp>

Inheritance diagram for cv::SparseMat:
Collaboration diagram for cv::SparseMat:

Classes

struct  Hdr
 the sparse matrix header More...
 
struct  Node
 sparse matrix node - element of a hash table More...
 

Public Types

enum  {
  MAGIC_VAL =0x42FD0000,
  MAX_DIM =32,
  HASH_SCALE =0x5bd1e995,
  HASH_BIT =0x80000000
}
 
typedef SparseMatConstIterator const_iterator
 
typedef SparseMatIterator iterator
 

Public Member Functions

 SparseMat ()
 Various SparseMat constructors. More...
 
 SparseMat (int dims, const int *_sizes, int _type)
 This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts. More...
 
 SparseMat (const SparseMat &m)
 This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts. More...
 
 SparseMat (const Mat &m)
 This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts. More...
 
 ~SparseMat ()
 the destructor More...
 
void addref ()
 manually increments the reference counter to the header. More...
 
void assignTo (SparseMat &m, int type=-1) const
 
int channels () const
 returns the number of channels More...
 
void clear ()
 sets all the sparse matrix elements to 0, which means clearing the hash table. More...
 
SparseMat clone () const CV_NODISCARD
 creates full copy of the matrix More...
 
void convertTo (SparseMat &m, int rtype, double alpha=1) const
 multiplies all the matrix elements by the specified scale factor alpha and converts the results to the specified data type More...
 
void convertTo (Mat &m, int rtype, double alpha=1, double beta=0) const
 converts sparse matrix to dense n-dim matrix with optional type conversion and scaling. More...
 
void copyTo (SparseMat &m) const
 copies all the data to the destination matrix. All the previous content of m is erased More...
 
void copyTo (Mat &m) const
 converts sparse matrix to dense matrix. More...
 
void create (int dims, const int *_sizes, int _type)
 reallocates sparse matrix. More...
 
int depth () const
 returns the depth of sparse matrix elements More...
 
int dims () const
 returns the matrix dimensionality More...
 
size_t elemSize () const
 converts sparse matrix to the old-style representation; all the elements are copied. More...
 
size_t elemSize1 () const
 returns elemSize()/channels() More...
 
SparseMatIterator end ()
 return the sparse matrix iterator pointing to the element following the last sparse matrix element More...
 
SparseMatConstIterator end () const
 returns the read-only sparse matrix iterator at the matrix end More...
 
template<typename _Tp >
SparseMatIterator_< _Tp > end ()
 returns the typed sparse matrix iterator at the matrix end More...
 
template<typename _Tp >
SparseMatConstIterator_< _Tp > end () const
 returns the typed read-only sparse matrix iterator at the matrix end More...
 
void erase (int i0, int i1, size_t *hashval=0)
 erases the specified element (2D case) More...
 
void erase (int i0, int i1, int i2, size_t *hashval=0)
 erases the specified element (3D case) More...
 
void erase (const int *idx, size_t *hashval=0)
 erases the specified element (nD case) More...
 
size_t hash (int i0) const
 computes the element hash value (1D case) More...
 
size_t hash (int i0, int i1) const
 computes the element hash value (2D case) More...
 
size_t hash (int i0, int i1, int i2) const
 computes the element hash value (3D case) More...
 
size_t hash (const int *idx) const
 computes the element hash value (nD case) More...
 
ucharnewNode (const int *idx, size_t hashval)
 
Nodenode (size_t nidx)
 
const Nodenode (size_t nidx) const
 
size_t nzcount () const
 returns the number of non-zero elements (=the number of hash table nodes) More...
 
SparseMatoperator= (const SparseMat &m)
 assignment operator. This is O(1) operation, i.e. no data is copied More...
 
SparseMatoperator= (const Mat &m)
 equivalent to the corresponding constructor More...
 
void release ()
 
void removeNode (size_t hidx, size_t nidx, size_t previdx)
 
void resizeHashTab (size_t newsize)
 
const int * size () const
 returns the array of sizes, or NULL if the matrix is not allocated More...
 
int size (int i) const
 returns the size of i-th matrix dimension (or 0) More...
 
int type () const
 returns type of sparse matrix elements More...
 
template<typename _Tp >
_Tp & value (Node *n)
 returns the value stored in the sparse martix node More...
 
template<typename _Tp >
const _Tp & value (const Node *n) const
 returns the value stored in the sparse martix node More...
 
ucharptr (int i0, bool createMissing, size_t *hashval=0)
 specialized variants for 1D, 2D, 3D cases and the generic_type one for n-D case. More...
 
ucharptr (int i0, int i1, bool createMissing, size_t *hashval=0)
 returns pointer to the specified element (2D case) More...
 
ucharptr (int i0, int i1, int i2, bool createMissing, size_t *hashval=0)
 returns pointer to the specified element (3D case) More...
 
ucharptr (const int *idx, bool createMissing, size_t *hashval=0)
 returns pointer to the specified element (nD case) More...
 
template<typename _Tp >
_Tp & ref (int i0, size_t *hashval=0)
 return read-write reference to the specified sparse matrix element. More...
 
template<typename _Tp >
_Tp & ref (int i0, int i1, size_t *hashval=0)
 returns reference to the specified element (2D case) More...
 
template<typename _Tp >
_Tp & ref (int i0, int i1, int i2, size_t *hashval=0)
 returns reference to the specified element (3D case) More...
 
template<typename _Tp >
_Tp & ref (const int *idx, size_t *hashval=0)
 returns reference to the specified element (nD case) More...
 
template<typename _Tp >
_Tp value (int i0, size_t *hashval=0) const
 return value of the specified sparse matrix element. More...
 
template<typename _Tp >
_Tp value (int i0, int i1, size_t *hashval=0) const
 returns value of the specified element (2D case) More...
 
template<typename _Tp >
_Tp value (int i0, int i1, int i2, size_t *hashval=0) const
 returns value of the specified element (3D case) More...
 
template<typename _Tp >
_Tp value (const int *idx, size_t *hashval=0) const
 returns value of the specified element (nD case) More...
 
template<typename _Tp >
const _Tp * find (int i0, size_t *hashval=0) const
 Return pointer to the specified sparse matrix element if it exists. More...
 
template<typename _Tp >
const _Tp * find (int i0, int i1, size_t *hashval=0) const
 returns pointer to the specified element (2D case) More...
 
template<typename _Tp >
const _Tp * find (int i0, int i1, int i2, size_t *hashval=0) const
 returns pointer to the specified element (3D case) More...
 
template<typename _Tp >
const _Tp * find (const int *idx, size_t *hashval=0) const
 returns pointer to the specified element (nD case) More...
 
SparseMatIterator begin ()
 return the sparse matrix iterator pointing to the first sparse matrix element More...
 
template<typename _Tp >
SparseMatIterator_< _Tp > begin ()
 returns the sparse matrix iterator at the matrix beginning More...
 
SparseMatConstIterator begin () const
 returns the read-only sparse matrix iterator at the matrix beginning More...
 
template<typename _Tp >
SparseMatConstIterator_< _Tp > begin () const
 returns the read-only sparse matrix iterator at the matrix beginning More...
 

Public Attributes

int flags
 
Hdrhdr
 

Detailed Description

The class SparseMat represents multi-dimensional sparse numerical arrays.

Such a sparse array can store elements of any type that Mat can store. Sparse means that only non-zero elements are stored (though, as a result of operations on a sparse matrix, some of its stored elements can actually become 0. It is up to you to detect such elements and delete them using SparseMat::erase ). The non-zero elements are stored in a hash table that grows when it is filled so that the search time is O(1) in average (regardless of whether element is there or not). Elements can be accessed using the following methods:

  • Query operations (SparseMat::ptr and the higher-level SparseMat::ref, SparseMat::value and SparseMat::find), for example:
    const int dims = 5;
    int size[5] = {10, 10, 10, 10, 10};
    SparseMat sparse_mat(dims, size, CV_32F);
    for(int i = 0; i < 1000; i++)
    {
    int idx[dims];
    for(int k = 0; k < dims; k++)
    idx[k] = rand() % size[k];
    sparse_mat.ref<float>(idx) += 1.f;
    }
    cout << "nnz = " << sparse_mat.nzcount() << endl;
  • Sparse matrix iterators. They are similar to MatIterator but different from NAryMatIterator. That is, the iteration loop is familiar to STL users:
    // prints elements of a sparse floating-point matrix
    // and the sum of elements.
    SparseMatConstIterator_<float>
    it = sparse_mat.begin<float>(),
    it_end = sparse_mat.end<float>();
    double s = 0;
    int dims = sparse_mat.dims();
    for(; it != it_end; ++it)
    {
    // print element indices and the element value
    const SparseMat::Node* n = it.node();
    printf("(");
    for(int i = 0; i < dims; i++)
    printf("%d%s", n->idx[i], i < dims-1 ? ", " : ")");
    printf(": %g\n", it.value<float>());
    s += *it;
    }
    printf("Element sum is %g\n", s);
    If you run this loop, you will notice that elements are not enumerated in a logical order (lexicographical, and so on). They come in the same order as they are stored in the hash table (semi-randomly). You may collect pointers to the nodes and sort them to get the proper ordering. Note, however, that pointers to the nodes may become invalid when you add more elements to the matrix. This may happen due to possible buffer reallocation.
  • Combination of the above 2 methods when you need to process 2 or more sparse matrices simultaneously. For example, this is how you can compute unnormalized cross-correlation of the 2 floating-point sparse matrices:
    double cross_corr(const SparseMat& a, const SparseMat& b)
    {
    const SparseMat *_a = &a, *_b = &b;
    // if b contains less elements than a,
    // it is faster to iterate through b
    if(_a->nzcount() > _b->nzcount())
    std::swap(_a, _b);
    SparseMatConstIterator_<float> it = _a->begin<float>(),
    it_end = _a->end<float>();
    double ccorr = 0;
    for(; it != it_end; ++it)
    {
    // take the next element from the first matrix
    float avalue = *it;
    const Node* anode = it.node();
    // and try to find an element with the same index in the second matrix.
    // since the hash value depends only on the element index,
    // reuse the hash value stored in the node
    float bvalue = _b->value<float>(anode->idx,&anode->hashval);
    ccorr += avalue*bvalue;
    }
    return ccorr;
    }

Member Typedef Documentation

◆ const_iterator

◆ iterator

Member Enumeration Documentation

◆ anonymous enum

anonymous enum
Enumerator
MAGIC_VAL 
MAX_DIM 
HASH_SCALE 
HASH_BIT 

Constructor & Destructor Documentation

◆ SparseMat() [1/4]

cv::SparseMat::SparseMat ( )

Various SparseMat constructors.

◆ SparseMat() [2/4]

cv::SparseMat::SparseMat ( int  dims,
const int *  _sizes,
int  _type 
)

This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

Parameters
dimsArray dimensionality.
_sizesSparce matrix size on all dementions.
_typeSparse matrix data type.

◆ SparseMat() [3/4]

cv::SparseMat::SparseMat ( const SparseMat m)

This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

Parameters
mSource matrix for copy constructor. If m is dense matrix (ocvMat) then it will be converted to sparse representation.

◆ SparseMat() [4/4]

cv::SparseMat::SparseMat ( const Mat m)
explicit

This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

Parameters
mSource matrix for copy constructor. If m is dense matrix (ocvMat) then it will be converted to sparse representation.

◆ ~SparseMat()

cv::SparseMat::~SparseMat ( )

the destructor

Member Function Documentation

◆ addref()

void cv::SparseMat::addref ( )

manually increments the reference counter to the header.

◆ assignTo()

void cv::SparseMat::assignTo ( SparseMat m,
int  type = -1 
) const

◆ begin() [1/4]

SparseMatIterator cv::SparseMat::begin ( )

return the sparse matrix iterator pointing to the first sparse matrix element

returns the sparse matrix iterator at the matrix beginning

◆ begin() [2/4]

template<typename _Tp >
SparseMatIterator_<_Tp> cv::SparseMat::begin ( )

returns the sparse matrix iterator at the matrix beginning

◆ begin() [3/4]

SparseMatConstIterator cv::SparseMat::begin ( ) const

returns the read-only sparse matrix iterator at the matrix beginning

◆ begin() [4/4]

template<typename _Tp >
SparseMatConstIterator_<_Tp> cv::SparseMat::begin ( ) const

returns the read-only sparse matrix iterator at the matrix beginning

◆ channels()

int cv::SparseMat::channels ( ) const

returns the number of channels

◆ clear()

void cv::SparseMat::clear ( )

sets all the sparse matrix elements to 0, which means clearing the hash table.

◆ clone()

SparseMat cv::SparseMat::clone ( ) const

creates full copy of the matrix

◆ convertTo() [1/2]

void cv::SparseMat::convertTo ( SparseMat m,
int  rtype,
double  alpha = 1 
) const

multiplies all the matrix elements by the specified scale factor alpha and converts the results to the specified data type

◆ convertTo() [2/2]

void cv::SparseMat::convertTo ( Mat m,
int  rtype,
double  alpha = 1,
double  beta = 0 
) const

converts sparse matrix to dense n-dim matrix with optional type conversion and scaling.

Parameters
[out]m- output matrix; if it does not have a proper size or type before the operation, it is reallocated
[in]rtype- desired output matrix type or, rather, the depth since the number of channels are the same as the input has; if rtype is negative, the output matrix will have the same type as the input.
[in]alpha- optional scale factor
[in]beta- optional delta added to the scaled values

◆ copyTo() [1/2]

void cv::SparseMat::copyTo ( SparseMat m) const

copies all the data to the destination matrix. All the previous content of m is erased

◆ copyTo() [2/2]

void cv::SparseMat::copyTo ( Mat m) const

converts sparse matrix to dense matrix.

◆ create()

void cv::SparseMat::create ( int  dims,
const int *  _sizes,
int  _type 
)

reallocates sparse matrix.

If the matrix already had the proper size and type, it is simply cleared with clear(), otherwise, the old matrix is released (using release()) and the new one is allocated.

◆ depth()

int cv::SparseMat::depth ( ) const

returns the depth of sparse matrix elements

◆ dims()

int cv::SparseMat::dims ( ) const

returns the matrix dimensionality

◆ elemSize()

size_t cv::SparseMat::elemSize ( ) const

converts sparse matrix to the old-style representation; all the elements are copied.

returns the size of each element in bytes (not including the overhead - the space occupied by SparseMat::Node elements)

◆ elemSize1()

size_t cv::SparseMat::elemSize1 ( ) const

returns elemSize()/channels()

◆ end() [1/4]

SparseMatIterator cv::SparseMat::end ( )

return the sparse matrix iterator pointing to the element following the last sparse matrix element

returns the sparse matrix iterator at the matrix end

◆ end() [2/4]

SparseMatConstIterator cv::SparseMat::end ( ) const

returns the read-only sparse matrix iterator at the matrix end

◆ end() [3/4]

template<typename _Tp >
SparseMatIterator_<_Tp> cv::SparseMat::end ( )

returns the typed sparse matrix iterator at the matrix end

◆ end() [4/4]

template<typename _Tp >
SparseMatConstIterator_<_Tp> cv::SparseMat::end ( ) const

returns the typed read-only sparse matrix iterator at the matrix end

◆ erase() [1/3]

void cv::SparseMat::erase ( int  i0,
int  i1,
size_t *  hashval = 0 
)

erases the specified element (2D case)

◆ erase() [2/3]

void cv::SparseMat::erase ( int  i0,
int  i1,
int  i2,
size_t *  hashval = 0 
)

erases the specified element (3D case)

◆ erase() [3/3]

void cv::SparseMat::erase ( const int *  idx,
size_t *  hashval = 0 
)

erases the specified element (nD case)

◆ find() [1/4]

template<typename _Tp >
const _Tp* cv::SparseMat::find ( int  i0,
size_t *  hashval = 0 
) const

Return pointer to the specified sparse matrix element if it exists.

find<_Tp>(i0,...[,hashval]) is equivalent to (_const Tp*)ptr(i0,...false[,hashval]).

If the specified element does not exist, the methods return NULL.returns pointer to the specified element (1D case)

◆ find() [2/4]

template<typename _Tp >
const _Tp* cv::SparseMat::find ( int  i0,
int  i1,
size_t *  hashval = 0 
) const

returns pointer to the specified element (2D case)

◆ find() [3/4]

template<typename _Tp >
const _Tp* cv::SparseMat::find ( int  i0,
int  i1,
int  i2,
size_t *  hashval = 0 
) const

returns pointer to the specified element (3D case)

◆ find() [4/4]

template<typename _Tp >
const _Tp* cv::SparseMat::find ( const int *  idx,
size_t *  hashval = 0 
) const

returns pointer to the specified element (nD case)

◆ hash() [1/4]

size_t cv::SparseMat::hash ( int  i0) const

computes the element hash value (1D case)

◆ hash() [2/4]

size_t cv::SparseMat::hash ( int  i0,
int  i1 
) const

computes the element hash value (2D case)

◆ hash() [3/4]

size_t cv::SparseMat::hash ( int  i0,
int  i1,
int  i2 
) const

computes the element hash value (3D case)

◆ hash() [4/4]

size_t cv::SparseMat::hash ( const int *  idx) const

computes the element hash value (nD case)

◆ newNode()

uchar* cv::SparseMat::newNode ( const int *  idx,
size_t  hashval 
)

◆ node() [1/2]

Node* cv::SparseMat::node ( size_t  nidx)

◆ node() [2/2]

const Node* cv::SparseMat::node ( size_t  nidx) const

◆ nzcount()

size_t cv::SparseMat::nzcount ( ) const

returns the number of non-zero elements (=the number of hash table nodes)

◆ operator=() [1/2]

SparseMat& cv::SparseMat::operator= ( const SparseMat m)

assignment operator. This is O(1) operation, i.e. no data is copied

◆ operator=() [2/2]

SparseMat& cv::SparseMat::operator= ( const Mat m)

equivalent to the corresponding constructor

◆ ptr() [1/4]

uchar* cv::SparseMat::ptr ( int  i0,
bool  createMissing,
size_t *  hashval = 0 
)

specialized variants for 1D, 2D, 3D cases and the generic_type one for n-D case.

return pointer to the matrix element.

  • if the element is there (it's non-zero), the pointer to it is returned
  • if it's not there and createMissing=false, NULL pointer is returned
  • if it's not there and createMissing=true, then the new element is created and initialized with 0. Pointer to it is returned
  • if the optional hashval pointer is not NULL, the element hash value is not computed, but *hashval is taken instead.returns pointer to the specified element (1D case)

◆ ptr() [2/4]

uchar* cv::SparseMat::ptr ( int  i0,
int  i1,
bool  createMissing,
size_t *  hashval = 0 
)

returns pointer to the specified element (2D case)

◆ ptr() [3/4]

uchar* cv::SparseMat::ptr ( int  i0,
int  i1,
int  i2,
bool  createMissing,
size_t *  hashval = 0 
)

returns pointer to the specified element (3D case)

◆ ptr() [4/4]

uchar* cv::SparseMat::ptr ( const int *  idx,
bool  createMissing,
size_t *  hashval = 0 
)

returns pointer to the specified element (nD case)

◆ ref() [1/4]

template<typename _Tp >
_Tp& cv::SparseMat::ref ( int  i0,
size_t *  hashval = 0 
)

return read-write reference to the specified sparse matrix element.

ref<_Tp>(i0,...[,hashval]) is equivalent to *(_Tp*)ptr(i0,...,true[,hashval]). The methods always return a valid reference. If the element did not exist, it is created and initialiazed with 0.returns reference to the specified element (1D case)

◆ ref() [2/4]

template<typename _Tp >
_Tp& cv::SparseMat::ref ( int  i0,
int  i1,
size_t *  hashval = 0 
)

returns reference to the specified element (2D case)

◆ ref() [3/4]

template<typename _Tp >
_Tp& cv::SparseMat::ref ( int  i0,
int  i1,
int  i2,
size_t *  hashval = 0 
)

returns reference to the specified element (3D case)

◆ ref() [4/4]

template<typename _Tp >
_Tp& cv::SparseMat::ref ( const int *  idx,
size_t *  hashval = 0 
)

returns reference to the specified element (nD case)

◆ release()

void cv::SparseMat::release ( )

◆ removeNode()

void cv::SparseMat::removeNode ( size_t  hidx,
size_t  nidx,
size_t  previdx 
)

◆ resizeHashTab()

void cv::SparseMat::resizeHashTab ( size_t  newsize)

◆ size() [1/2]

const int* cv::SparseMat::size ( ) const

returns the array of sizes, or NULL if the matrix is not allocated

◆ size() [2/2]

int cv::SparseMat::size ( int  i) const

returns the size of i-th matrix dimension (or 0)

◆ type()

int cv::SparseMat::type ( ) const

returns type of sparse matrix elements

◆ value() [1/6]

template<typename _Tp >
_Tp cv::SparseMat::value ( int  i0,
size_t *  hashval = 0 
) const

return value of the specified sparse matrix element.

value<_Tp>(i0,...[,hashval]) is equivalent to

{ const _Tp* p = find<_Tp>(i0,...[,hashval]); return p ? *p : _Tp(); }

That is, if the element did not exist, the methods return 0.returns value of the specified element (1D case)

◆ value() [2/6]

template<typename _Tp >
_Tp cv::SparseMat::value ( int  i0,
int  i1,
size_t *  hashval = 0 
) const

returns value of the specified element (2D case)

◆ value() [3/6]

template<typename _Tp >
_Tp cv::SparseMat::value ( int  i0,
int  i1,
int  i2,
size_t *  hashval = 0 
) const

returns value of the specified element (3D case)

◆ value() [4/6]

template<typename _Tp >
_Tp cv::SparseMat::value ( const int *  idx,
size_t *  hashval = 0 
) const

returns value of the specified element (nD case)

◆ value() [5/6]

template<typename _Tp >
_Tp& cv::SparseMat::value ( Node n)

returns the value stored in the sparse martix node

◆ value() [6/6]

template<typename _Tp >
const _Tp& cv::SparseMat::value ( const Node n) const

returns the value stored in the sparse martix node

Member Data Documentation

◆ flags

int cv::SparseMat::flags

◆ hdr

Hdr* cv::SparseMat::hdr

The documentation for this class was generated from the following file: