Learning OpenCV, 2nd Edition

}

In this example, we also slipped in the method node(), which is defined for the iterators. node()

returns a pointer to the internal data node in the sparse matrix that is indicated by the iterator. The returned

object of type cv::SparseMat::Node has the following definition:

struct Node

{

size_t hashval;

size_t next;

int idx[cv::MAX_DIM];

};

This structure contains both the index of the associated element (note that element idx is of type int[]),

as well as the hash value associated with that node (element hashval is the same hash value as can be

used with SparseMat::ptr(), SparseMat::ref(), SparseMat::value(), and

SparseMat::find().)

Functions Unique to Sparse Arrays

As stated earlier, sparse matrices support many of the same operations as dense matrices. In addition, there

are several methods that are unique to sparse matrices. These are listed in Table 3-19, and include the

functions mentioned in the previous sections.

Table 3-19: Additional class member functions of cv::SparseMat

Example

Description

cv::SparseMat sm();

Create a sparse matrix without initialization

cv::SparseMat sm( 3, sz, cv::F32 );

Create a three-dimensional sparse matrix with

dimensions given by the array sz of type float

cv::SparseMat sm( sm0 );

Create a new sparse matrix which is a copy of

existing sparse matrix sm0

cv::SparseMat( m0, try1d );

Create a spa