Skip to content

O
opencv_contrib
Commit 32aad039 authored by KUANG Fangjun's avatar KUANG Fangjun
Browse files
Options

Improve the HDF module. 

 - Some test cases are added
 - Two tutorials are written to show how to use it.
parent 53986aaa
Hide whitespace changes
Inline Side-by-side
Showing with 802 additions and 174 deletions
modules/fuzzy/samples/fuzzy_filtering.cpp
View file @ 32aad039
......@@ -11,7 +11,7 @@
* using "kernel2" with radius 100.
*
* Both kernels are created from linear function, using
* linear interpolation (parametr ft:LINEAR).
* linear interpolation (parameter ft:LINEAR).
*/
#include "opencv2/core.hpp"
......@@ -26,7 +26,7 @@ int main(void)
// Input image
Mat I = imread("input.png");
// Kernel cretion
// Kernel creation
Mat kernel1, kernel2;
ft::createKernel(ft::LINEAR, 3, kernel1, 3);
......
modules/hdf/include/opencv2/hdf.hpp
View file @ 32aad039
......@@ -47,6 +47,9 @@ This module provides storage routines for Hierarchical Data Format objects.
Hierarchical Data Format version 5
--------------------------------------------------------
In order to use it, the hdf5 library has to be installed, which
means cmake should find it using `find_package(HDF5)` .
@}
*/
......
modules/hdf/include/opencv2/hdf/hdf5.hpp
View file @ 32aad039
......@@ -36,7 +36,7 @@
#define __OPENCV_HDF5_HPP__
#include <vector>
#include <opencv2/core.hpp>
namespace cv
{
......@@ -50,7 +50,7 @@ using namespace std;
/** @brief Hierarchical Data Format version 5 interface.
Notice that module is compiled only when hdf5 is correctly installed.
Notice that this module is compiled only when hdf5 is correctly installed.
*/
class CV_EXPORTS_W HDF5
......@@ -59,7 +59,11 @@ public:
CV_WRAP enum
{
H5_UNLIMITED = -1, H5_NONE = -1, H5_GETDIMS = 100, H5_GETMAXDIMS = 101, H5_GETCHUNKDIMS = 102,
H5_UNLIMITED = -1, //!< The dimension size is unlimited, @sa dscreate()
H5_NONE = -1, //!< No compression, @sa dscreate()
H5_GETDIMS = 100, //!< Get the dimension information of a dataset. @sa dsgetsize()
H5_GETMAXDIMS = 101, //!< Get the maximum dimension information of a dataset. @sa dsgetsize()
H5_GETCHUNKDIMS = 102, //!< Get the chunk sizes of a dataset. @sa dsgetsize()
};
virtual ~HDF5() {}
......@@ -71,63 +75,57 @@ public:
/** @brief Create a group.
@param grlabel specify the hdf5 group label.
Create a hdf5 group.
Create a hdf5 group with default properties. The group is closed automatically after creation.
@note Groups are useful for better organise multiple datasets. It is possible to create subgroups within any group.
Existence of a particular group can be checked using hlexists(). In case of subgroups label would be e.g: 'Group1/SubGroup1'
where SubGroup1 is within the root group Group1.
@note Groups are useful for better organising multiple datasets. It is possible to create subgroups within any group.
Existence of a particular group can be checked using hlexists(). In case of subgroups, a label would be e.g: 'Group1/SubGroup1'
where SubGroup1 is within the root group Group1. Before creating a subgroup, its parent group MUST be created.
- In this example Group1 will have one subgrup labeled SubGroup1:
@code{.cpp}
// open / autocreate hdf5 file
cv::Ptr<cv::hdf::HDF5> h5io = cv::hdf::open( "mytest.h5" );
// create Group1 if does not exists
if ( ! h5io->hlexists( "Group1" ) )
h5io->grcreate( "Group1" );
else
printf("Group1 already created, skipping\n" );
// create SubGroup1 if does not exists
if ( ! h5io->hlexists( "Group1/SubGroup1" ) )
h5io->grcreate( "Group1/SubGroup1" );
else
printf("SubGroup1 already created, skipping\n" );
// release
h5io->close();
@endcode
- In this example, Group1 will have one subgroup called SubGroup1:
@note When a dataset is created with dscreate() or kpcreate() it can be created right within a group by specifying
full path within the label, in our example would be: 'Group1/SubGroup1/MyDataSet'. It is not thread safe.
@snippet samples/create_groups.cpp create_group
The corresponding result visualized using the HDFView tool is
![Visualization of groups using the HDFView tool](pics/create_groups.png)
@note When a dataset is created with dscreate() or kpcreate(), it can be created within a group by specifying the
full path within the label. In our example, it would be: 'Group1/SubGroup1/MyDataSet'. It is not thread safe.
*/
CV_WRAP virtual void grcreate( String grlabel ) = 0;
CV_WRAP virtual void grcreate( const String& grlabel ) = 0;
/** @brief Check if label exists or not.
@param label specify the hdf5 dataset label.
Returns **true** if dataset exists, and **false** if does not.
Returns **true** if dataset exists, and **false** otherwise.
@note Checks if dataset, group or other object type (hdf5 link) exists under the label name. It is thread safe.
*/
CV_WRAP virtual bool hlexists( String label ) const = 0;
CV_WRAP virtual bool hlexists( const String& label ) const = 0;
/* @overload */
CV_WRAP virtual void dscreate( const int rows, const int cols, const int type,
String dslabel ) const = 0;
const String& dslabel ) const = 0;
/* @overload */
CV_WRAP virtual void dscreate( const int rows, const int cols, const int type,
String dslabel, const int compresslevel ) const = 0;
const String& dslabel, const int compresslevel ) const = 0;
/* @overload */
CV_WRAP virtual void dscreate( const int rows, const int cols, const int type,
String dslabel, const int compresslevel, const vector<int>& dims_chunks ) const = 0;
const String& dslabel, const int compresslevel, const vector<int>& dims_chunks ) const = 0;
/** @brief Create and allocate storage for two dimensional single or multi channel dataset.
@param rows declare amount of rows
@param cols declare amount of cols
@param type type to be used
@param dslabel specify the hdf5 dataset label, any existing dataset with the same label will be overwritten.
@param compresslevel specify the compression level 0-9 to be used, H5_NONE is default and means no compression.
@param dims_chunks each array member specify chunking sizes to be used for block i/o,
@param cols declare amount of columns
@param type type to be used, e.g, CV_8UC3, CV_32FC1 and etc.
@param dslabel specify the hdf5 dataset label. Existing dataset label will cause an error.
@param compresslevel specify the compression level 0-9 to be used, H5_NONE is the default value and means no compression.
The value 0 also means no compression.
A value 9 indicating the best compression ration. Note
that a higher compression level indicates a higher computational cost. It relies
on GNU gzip for compression.
@param dims_chunks each array member specifies the chunking size to be used for block I/O,
by default NULL means none at all.
@note If the dataset already exists an exception will be thrown.
@note If the dataset already exists, an exception will be thrown (CV_Error() is called).
- Existence of the dataset can be checked using hlexists(), see in this example:
@code{.cpp}
......@@ -143,9 +141,9 @@ public:
@endcode
@note Activating compression requires internal chunking. Chunking can significantly improve access
speed booth at read or write time especially for windowed access logic that shifts offset inside dataset.
If no custom chunking is specified default one will be invoked by the size of **whole** dataset
as single big chunk of data.
speed both at read and write time, especially for windowed access logic that shifts offset inside dataset.
If no custom chunking is specified, the default one will be invoked by the size of the **whole** dataset
as a single big chunk of data.
- See example of level 9 compression using internal default chunking:
@code{.cpp}
......@@ -160,11 +158,11 @@ public:
h5io->close();
@endcode
@note A value of H5_UNLIMITED for **rows** or **cols** or booth means **unlimited** data on the specified dimension,
thus is possible to expand anytime such dataset on row, col or booth directions. Presence of H5_UNLIMITED on any
dimension **require** to define custom chunking. No default chunking will be defined in unlimited scenario since
default size on that dimension will be zero, and will grow once dataset is written. Writing into dataset that have
H5_UNLIMITED on some of its dimension requires dsinsert() that allow growth on unlimited dimension instead of dswrite()
@note A value of H5_UNLIMITED for **rows** or **cols** or both means **unlimited** data on the specified dimension,
thus, it is possible to expand anytime such a dataset on row, col or on both directions. Presence of H5_UNLIMITED on any
dimension **requires** to define custom chunking. No default chunking will be defined in the unlimited scenario since
default size on that dimension will be zero, and will grow once dataset is written. Writing into a dataset that has
H5_UNLIMITED on some of its dimensions requires dsinsert() that allows growth on unlimited dimensions, instead of dswrite()
that allows to write only in predefined data space.
- Example below shows no compression but unlimited dimension on cols using 100x100 internal chunking:
......@@ -178,31 +176,35 @@ public:
h5io->close();
@endcode
@note It is **not** thread safe, it must be called only once at dataset creation otherwise exception will occur.
Multiple datasets inside single hdf5 file is allowed.
@note It is **not** thread safe, it must be called only once at dataset creation, otherwise an exception will occur.
Multiple datasets inside a single hdf5 file are allowed.
*/
CV_WRAP virtual void dscreate( const int rows, const int cols, const int type,
String dslabel, const int compresslevel, const int* dims_chunks ) const = 0;
const String& dslabel, const int compresslevel, const int* dims_chunks ) const = 0;
/* @overload */
CV_WRAP virtual void dscreate( const int n_dims, const int* sizes, const int type,
String dslabel ) const = 0;
const String& dslabel ) const = 0;
/* @overload */
CV_WRAP virtual void dscreate( const int n_dims, const int* sizes, const int type,
String dslabel, const int compresslevel ) const = 0;
const String& dslabel, const int compresslevel ) const = 0;
/* @overload */
CV_WRAP virtual void dscreate( const vector<int>& sizes, const int type,
String dslabel, const int compresslevel = HDF5::H5_NONE,
const String& dslabel, const int compresslevel = HDF5::H5_NONE,
const vector<int>& dims_chunks = vector<int>() ) const = 0;
/** @brief Create and allocate storage for n-dimensional dataset, single or mutichannel type.
/** @brief Create and allocate storage for n-dimensional dataset, single or multichannel type.
@param n_dims declare number of dimensions
@param sizes array containing sizes for each dimensions
@param type type to be used
@param dslabel specify the hdf5 dataset label, any existing dataset with the same label will be overwritten.
@param compresslevel specify the compression level 0-9 to be used, H5_NONE is default and means no compression.
@param dims_chunks each array member specify chunking sizes to be used for block i/o,
@param type type to be used, e.g., CV_8UC3, CV_32FC1, etc.
@param dslabel specify the hdf5 dataset label. Existing dataset label will cause an error.
@param compresslevel specify the compression level 0-9 to be used, H5_NONE is the default value and means no compression.
The value 0 also means no compression.
A value 9 indicating the best compression ration. Note
that a higher compression level indicates a higher computational cost. It relies
on GNU gzip for compression.
@param dims_chunks each array member specifies chunking sizes to be used for block I/O,
by default NULL means none at all.
@note If the dataset already exists an exception will be thrown. Existence of the dataset can be checked
@note If the dataset already exists, an exception will be thrown. Existence of the dataset can be checked
using hlexists().
- See example below that creates a 6 dimensional storage space:
......@@ -221,12 +223,12 @@ public:
@endcode
@note Activating compression requires internal chunking. Chunking can significantly improve access
speed booth at read or write time especially for windowed access logic that shifts offset inside dataset.
If no custom chunking is specified default one will be invoked by the size of **whole** dataset
speed both at read and write time, especially for windowed access logic that shifts offset inside dataset.
If no custom chunking is specified, the default one will be invoked by the size of **whole** dataset
as single big chunk of data.
- See example of level 0 compression (shallow) using chunking against first
dimension, thus storage will consists by 100 chunks of data:
- See example of level 0 compression (shallow) using chunking against the first
dimension, thus storage will consists of 100 chunks of data:
@code{.cpp}
// open / autocreate hdf5 file
cv::Ptr<cv::hdf::HDF5> h5io = cv::hdf::open( "mytest.h5" );
......@@ -242,11 +244,11 @@ public:
h5io->close();
@endcode
@note A value of H5_UNLIMITED inside the **sizes** array means **unlimited** data on that dimension, thus is
@note A value of H5_UNLIMITED inside the **sizes** array means **unlimited** data on that dimension, thus it is
possible to expand anytime such dataset on those unlimited directions. Presence of H5_UNLIMITED on any dimension
**require** to define custom chunking. No default chunking will be defined in unlimited scenario since default size
on that dimension will be zero, and will grow once dataset is written. Writing into dataset that have H5_UNLIMITED on
some of its dimension requires dsinsert() instead of dswrite() that allow growth on unlimited dimension instead of
**requires** to define custom chunking. No default chunking will be defined in unlimited scenario since the default size
on that dimension will be zero, and will grow once dataset is written. Writing into dataset that has H5_UNLIMITED on
some of its dimension requires dsinsert() instead of dswrite() that allows growth on unlimited dimension instead of
dswrite() that allows to write only in predefined data space.
- Example below shows a 3 dimensional dataset using no compression with all unlimited sizes and one unit chunking:
......@@ -262,11 +264,12 @@ public:
@endcode
*/
CV_WRAP virtual void dscreate( const int n_dims, const int* sizes, const int type,
String dslabel, const int compresslevel, const int* dims_chunks ) const = 0;
const String& dslabel, const int compresslevel, const int* dims_chunks ) const = 0;
/** @brief Fetch dataset sizes
@param dslabel specify the hdf5 dataset label to be measured.
@param dims_flag will fetch dataset dimensions on H5_GETDIMS, and dataset maximum dimensions on H5_GETMAXDIMS.
@param dims_flag will fetch dataset dimensions on H5_GETDIMS, dataset maximum dimensions on H5_GETMAXDIMS,
and chunk sizes on H5_GETCHUNKDIMS.
Returns vector object containing sizes of dataset on each dimensions.
......@@ -278,7 +281,7 @@ public:
return the dimension of chunk if dataset was created with chunking options otherwise returned vector size
will be zero.
*/
CV_WRAP virtual vector<int> dsgetsize( String dslabel, int dims_flag = HDF5::H5_GETDIMS ) const = 0;
CV_WRAP virtual vector<int> dsgetsize( const String& dslabel, int dims_flag = HDF5::H5_GETDIMS ) const = 0;
/** @brief Fetch dataset type
@param dslabel specify the hdf5 dataset label to be checked.
......@@ -289,15 +292,15 @@ public:
@note Result can be parsed with CV_MAT_CN() to obtain amount of channels and CV_MAT_DEPTH() to obtain native cvdata type.
It is thread safe.
*/
CV_WRAP virtual int dsgettype( String dslabel ) const = 0;
CV_WRAP virtual int dsgettype( const String& dslabel ) const = 0;
/* @overload */
CV_WRAP virtual void dswrite( InputArray Array, String dslabel ) const = 0;
CV_WRAP virtual void dswrite( InputArray Array, const String& dslabel ) const = 0;
/* @overload */
CV_WRAP virtual void dswrite( InputArray Array, String dslabel,
CV_WRAP virtual void dswrite( InputArray Array, const String& dslabel,
const int* dims_offset ) const = 0;
/* @overload */
CV_WRAP virtual void dswrite( InputArray Array, String dslabel,
CV_WRAP virtual void dswrite( InputArray Array, const String& dslabel,
const vector<int>& dims_offset,
const vector<int>& dims_counts = vector<int>() ) const = 0;
/** @brief Write or overwrite a Mat object into specified dataset of hdf5 file.
......@@ -305,16 +308,16 @@ public:
@param dslabel specify the target hdf5 dataset label.
@param dims_offset each array member specify the offset location
over dataset's each dimensions from where InputArray will be (over)written into dataset.
@param dims_counts each array member specify the amount of data over dataset's
@param dims_counts each array member specifies the amount of data over dataset's
each dimensions from InputArray that will be written into dataset.
Writes Mat object into targeted dataset.
@note If dataset is not created and does not exist it will be created **automatically**. Only Mat is supported and
it must to be **continuous**. It is thread safe but it is recommended that writes to happen over separate non overlapping
regions. Multiple datasets can be written inside single hdf5 file.
it must be **continuous**. It is thread safe but it is recommended that writes to happen over separate non-overlapping
regions. Multiple datasets can be written inside a single hdf5 file.
- Example below writes a 100x100 CV_64FC2 matrix into a dataset. No dataset precreation required. If routine
- Example below writes a 100x100 CV_64FC2 matrix into a dataset. No dataset pre-creation required. If routine
is called multiple times dataset will be just overwritten:
@code{.cpp}
// dual channel hilbert matrix
......@@ -362,19 +365,19 @@ public:
h5io->close();
@endcode
*/
CV_WRAP virtual void dswrite( InputArray Array, String dslabel,
CV_WRAP virtual void dswrite( InputArray Array, const String& dslabel,
const int* dims_offset, const int* dims_counts ) const = 0;
/* @overload */
CV_WRAP virtual void dsinsert( InputArray Array, String dslabel ) const = 0;
CV_WRAP virtual void dsinsert( InputArray Array, const String& dslabel ) const = 0;
/* @overload */
CV_WRAP virtual void dsinsert( InputArray Array,
String dslabel, const int* dims_offset ) const = 0;
const String& dslabel, const int* dims_offset ) const = 0;
/* @overload */
CV_WRAP virtual void dsinsert( InputArray Array,
String dslabel, const vector<int>& dims_offset,
const String& dslabel, const vector<int>& dims_offset,
const vector<int>& dims_counts = vector<int>() ) const = 0;
/** @brief Insert or overwrite a Mat object into specified dataset and autoexpand dataset size if **unlimited** property allows.
/** @brief Insert or overwrite a Mat object into specified dataset and auto expand dataset size if **unlimited** property allows.
@param Array specify Mat data array to be written.
@param dslabel specify the target hdf5 dataset label.
@param dims_offset each array member specify the offset location
......@@ -384,13 +387,13 @@ public:
Writes Mat object into targeted dataset and **autoexpand** dataset dimension if allowed.
@note Unlike dswrite(), datasets are **not** created **automatically**. Only Mat is supported and it must to be **continuous**.
If dsinsert() happen over outer regions of dataset dimensions and on that dimension of dataset is in **unlimited** mode then
@note Unlike dswrite(), datasets are **not** created **automatically**. Only Mat is supported and it must be **continuous**.
If dsinsert() happens over outer regions of dataset dimensions and on that dimension of dataset is in **unlimited** mode then
dataset is expanded, otherwise exception is thrown. To create datasets with **unlimited** property on specific or more
dimensions see dscreate() and the optional H5_UNLIMITED flag at creation time. It is not thread safe over same dataset
but multiple datasets can be merged inside single hdf5 file.
but multiple datasets can be merged inside a single hdf5 file.
- Example below creates **unlimited** rows x 100 cols and expand rows 5 times with dsinsert() using single 100x100 CV_64FC2
- Example below creates **unlimited** rows x 100 cols and expands rows 5 times with dsinsert() using single 100x100 CV_64FC2
over the dataset. Final size will have 5x100 rows and 100 cols, reflecting H matrix five times over row's span. Chunks size is
100x100 just optimized against the H matrix size having compression disabled. If routine is called multiple times dataset will be
just overwritten:
......@@ -421,17 +424,17 @@ public:
h5io->close();
@endcode
*/
CV_WRAP virtual void dsinsert( InputArray Array, String dslabel,
CV_WRAP virtual void dsinsert( InputArray Array, const String& dslabel,
const int* dims_offset, const int* dims_counts ) const = 0;
/* @overload */
CV_WRAP virtual void dsread( OutputArray Array, String dslabel ) const = 0;
CV_WRAP virtual void dsread( OutputArray Array, const String& dslabel ) const = 0;
/* @overload */
CV_WRAP virtual void dsread( OutputArray Array,
String dslabel, const int* dims_offset ) const = 0;
const String& dslabel, const int* dims_offset ) const = 0;
/* @overload */
CV_WRAP virtual void dsread( OutputArray Array, String dslabel,
CV_WRAP virtual void dsread( OutputArray Array, const String& dslabel,
const vector<int>& dims_offset,
const vector<int>& dims_counts = vector<int>() ) const = 0;
/** @brief Read specific dataset from hdf5 file into Mat object.
......@@ -473,7 +476,7 @@ public:
h5io->close();
@endcode
*/
CV_WRAP virtual void dsread( OutputArray Array, String dslabel,
CV_WRAP virtual void dsread( OutputArray Array, const String& dslabel,
const int* dims_offset, const int* dims_counts ) const = 0;
/** @brief Fetch keypoint dataset size
......@@ -489,13 +492,13 @@ public:
exception. The H5_GETCHUNKDIMS will return the dimension of chunk if dataset was created with chunking options otherwise
returned vector size will be zero.
*/
CV_WRAP virtual int kpgetsize( String kplabel, int dims_flag = HDF5::H5_GETDIMS ) const = 0;
CV_WRAP virtual int kpgetsize( const String& kplabel, int dims_flag = HDF5::H5_GETDIMS ) const = 0;
/** @brief Create and allocate special storage for cv::KeyPoint dataset.
@param size declare fixed number of KeyPoints
@param kplabel specify the hdf5 dataset label, any existing dataset with the same label will be overwritten.
@param compresslevel specify the compression level 0-9 to be used, H5_NONE is default and means no compression.
@param chunks each array member specify chunking sizes to be used for block i/o,
@param chunks each array member specifies chunking sizes to be used for block I/O,
H5_NONE is default and means no compression.
@note If the dataset already exists an exception will be thrown. Existence of the dataset can be checked
using hlexists().
......@@ -526,7 +529,7 @@ public:
printf("DS already created, skipping\n" );
@endcode
*/
virtual void kpcreate( const int size, String kplabel,
virtual void kpcreate( const int size, const String& kplabel,
const int compresslevel = H5_NONE, const int chunks = H5_NONE ) const = 0;
/** @brief Write or overwrite list of KeyPoint into specified dataset of hdf5 file.
......@@ -579,7 +582,7 @@ public:
h5io->close();
@endcode
*/
virtual void kpwrite( const vector<KeyPoint> keypoints, String kplabel,
virtual void kpwrite( const vector<KeyPoint> keypoints, const String& kplabel,
const int offset = H5_NONE, const int counts = H5_NONE ) const = 0;
/** @brief Insert or overwrite list of KeyPoint into specified dataset and autoexpand dataset size if **unlimited** property allows.
......@@ -614,7 +617,7 @@ public:
h5io->close();
@endcode
*/
virtual void kpinsert( const vector<KeyPoint> keypoints, String kplabel,
virtual void kpinsert( const vector<KeyPoint> keypoints, const String& kplabel,
const int offset = H5_NONE, const int counts = H5_NONE ) const = 0;
/** @brief Read specific keypoint dataset from hdf5 file into vector<KeyPoint> object.
......@@ -652,7 +655,7 @@ public:
h5io->close();
@endcode
*/
virtual void kpread( vector<KeyPoint>& keypoints, String kplabel,
virtual void kpread( vector<KeyPoint>& keypoints, const String& kplabel,
const int offset = H5_NONE, const int counts = H5_NONE ) const = 0;
};
......@@ -660,18 +663,20 @@ public:
/** @brief Open or create hdf5 file
@param HDF5Filename specify the HDF5 filename.
Returns pointer to the hdf5 object class
Returns a pointer to the hdf5 object class
@note If hdf5 file does not exist it will be created. Any operations except dscreate() functions on object
will be thread safe. Multiple datasets can be created inside single hdf5 file, and can be accessed
from same hdf5 object from multiple instances as long read or write operations are done over
@note If the specified file does not exist, it will be created using default properties.
Otherwise, it is opened in read and write mode with default access properties.
Any operations except dscreate() functions on object
will be thread safe. Multiple datasets can be created inside a single hdf5 file, and can be accessed
from the same hdf5 object from multiple instances as long read or write operations are done over
non-overlapping regions of dataset. Single hdf5 file also can be opened by multiple instances,
reads and writes can be instantiated at the same time as long non-overlapping regions are involved. Object
reads and writes can be instantiated at the same time as long as non-overlapping regions are involved. Object
is released using close().
- Example below open and then release the file.
- Example below opens and then releases the file.
@code{.cpp}
// open / autocreate hdf5 file
// open / auto create hdf5 file
cv::Ptr<cv::hdf::HDF5> h5io = cv::hdf::open( "mytest.h5" );
// ...
// release
......@@ -698,7 +703,7 @@ public:
}
@endcode
*/
CV_EXPORTS_W Ptr<HDF5> open( String HDF5Filename );
CV_EXPORTS_W Ptr<HDF5> open( const String& HDF5Filename );
//! @}
......
modules/hdf/samples/create_groups.cpp 0 → 100644
View file @ 32aad039
// This file is part of OpenCV project.
// It is subject to the license terms in the LICENSE file found in the top-level directory
// of this distribution and at http://opencv.org/license.html.
/**
* @file create_groups.cpp
* @author Fangjun Kuang <csukuangfj dot at gmail dot com>
* @date December 2017
*
* @brief It demonstrates how to create HDF5 groups and subgroups.
*
* Basic steps:
* 1. Use hdf::open to create a HDF5 file
* 2. Use HDF5::hlexists to check if a group exists or not
* 3. Use HDF5::grcreate to create a group by specifying its name
* 4. Use hdf::close to close a HDF5 file after modifying it
*
*/
//! [tutorial]
#include <iostream>
#include <opencv2/core.hpp>
#include <opencv2/hdf.hpp>
using namespace cv;
int main()
{
//! [create_group]
//! [tutorial_create_file]
Ptr<hdf::HDF5> h5io = hdf::open("mytest.h5");
//! [tutorial_create_file]
//! [tutorial_create_group]
// "/" means the root group, which is always present
if (!h5io->hlexists("/Group1"))
h5io->grcreate("/Group1");
else
std::cout << "/Group1 has already been created, skip it.\n";
//! [tutorial_create_group]
//! [tutorial_create_subgroup]
// Note that Group1 has been created above, otherwise exception will occur
if (!h5io->hlexists("/Group1/SubGroup1"))
h5io->grcreate("/Group1/SubGroup1");
else
std::cout << "/Group1/SubGroup1 has already been created, skip it.\n";
//! [tutorial_create_subgroup]
//! [tutorial_close_file]
h5io->close();
//! [tutorial_close_file]
//! [create_group]
return 0;
}
//! [tutorial]
modules/hdf/samples/create_read_write_datasets.cpp 0 → 100644
View file @ 32aad039
// This file is part of OpenCV project.
// It is subject to the license terms in the LICENSE file found in the top-level directory
// of this distribution and at http://opencv.org/license.html.
/**
* @file create_read_write.cpp
* @author Fangjun Kuang <csukuangfj dot at gmail dot com>
* @date December 2017
*
* @brief It demonstrates how to create a dataset, how
* to write a cv::Mat to the dataset and how to
* read a cv::Mat from it.
*
*/
//! [tutorial]
#include <iostream>
#include <opencv2/core.hpp>
#include <opencv2/hdf.hpp>
using namespace cv;
void write_root_group_single_channel()
{
String filename = "root_group_single_channel.h5";
String dataset_name = "/single"; // Note that it is a child of the root group /
// prepare data
Mat data;
data = (cv::Mat_<float>(2, 3) << 0, 1, 2, 3, 4, 5, 6);
//! [tutorial_open_file]
Ptr<hdf::HDF5> h5io = hdf::open(filename);
//! [tutorial_open_file]
//! [tutorial_write_root_single_channel]
// write data to the given dataset
// the dataset "/single" is created automatically, since it is a child of the root
h5io->dswrite(data, dataset_name);
//! [tutorial_write_root_single_channel]
//! [tutorial_read_dataset]
Mat expected;
h5io->dsread(expected, dataset_name);
//! [tutorial_read_dataset]
//! [tutorial_check_result]
double diff = norm(data - expected);
CV_Assert(abs(diff) < 1e-10);
//! [tutorial_check_result]
h5io->close();
}
void write_single_channel()
{
String filename = "single_channel.h5";
String parent_name = "/data";
String dataset_name = parent_name + "/single";
// prepare data
Mat data;
data = (cv::Mat_<float>(2, 3) << 0, 1, 2, 3, 4, 5);
Ptr<hdf::HDF5> h5io = hdf::open(filename);
//! [tutorial_create_dataset]
// first we need to create the parent group
if (!h5io->hlexists(parent_name)) h5io->grcreate(parent_name);
// create the dataset if it not exists
if (!h5io->hlexists(dataset_name)) h5io->dscreate(data.rows, data.cols, data.type(), dataset_name);
//! [tutorial_create_dataset]
// the following is the same with the above function write_root_group_single_channel()
h5io->dswrite(data, dataset_name);
Mat expected;
h5io->dsread(expected, dataset_name);
double diff = norm(data - expected);
CV_Assert(abs(diff) < 1e-10);
h5io->close();
}
/*
* creating, reading and writing multiple-channel matrices
* are the same with single channel matrices
*/
void write_multiple_channels()
{
String filename = "two_channels.h5";
String parent_name = "/data";
String dataset_name = parent_name + "/two_channels";
// prepare data
Mat data(2, 3, CV_32SC2);
for (size_t i = 0; i < data.total()*data.channels(); i++)
((int*) data.data)[i] = (int)i;
Ptr<hdf::HDF5> h5io = hdf::open(filename);
// first we need to create the parent group
if (!h5io->hlexists(parent_name)) h5io->grcreate(parent_name);
// create the dataset if it not exists
if (!h5io->hlexists(dataset_name)) h5io->dscreate(data.rows, data.cols, data.type(), dataset_name);
// the following is the same with the above function write_root_group_single_channel()
h5io->dswrite(data, dataset_name);
Mat expected;
h5io->dsread(expected, dataset_name);
double diff = norm(data - expected);
CV_Assert(abs(diff) < 1e-10);
h5io->close();
}
int main()
{
write_root_group_single_channel();
write_single_channel();
write_multiple_channels();
return 0;
}
//! [tutorial]
modules/hdf/src/hdf5.cpp
View file @ 32aad039
......@@ -47,7 +47,7 @@ class HDF5Impl : public HDF5
{
public:
HDF5Impl( String HDF5Filename );
HDF5Impl( const String& HDF5Filename );
virtual ~HDF5Impl() { close(); };
......@@ -59,96 +59,96 @@ public:
*/
// check if object / link exists
virtual bool hlexists( String label ) const;
virtual bool hlexists( const String& label ) const;
/*
* h5 group
*/
// create a group
virtual void grcreate( String grlabel );
virtual void grcreate( const String& grlabel );
/*
* cv::Mat
*/
// get sizes of dataset
virtual vector<int> dsgetsize( String dslabel, int dims_flag = H5_GETDIMS ) const;
virtual vector<int> dsgetsize( const String& dslabel, int dims_flag = H5_GETDIMS ) const;
/* get data type of dataset */
virtual int dsgettype( String dslabel ) const;
virtual int dsgettype( const String& dslabel ) const;
// overload dscreate() #1
virtual void dscreate( const int rows, const int cols, const int type, String dslabel ) const;
virtual void dscreate( const int rows, const int cols, const int type, const String& dslabel ) const;
// overload dscreate() #2
virtual void dscreate( const int rows, const int cols, const int type, String dslabel,
virtual void dscreate( const int rows, const int cols, const int type, const String& dslabel,
const int compresslevel ) const;
// overload dscreate() #3
virtual void dscreate( const int rows, const int cols, const int type, String dslabel,
virtual void dscreate( const int rows, const int cols, const int type, const String& dslabel,
const int compresslevel, const vector<int>& dims_chunks ) const;
/* create two dimensional single or mutichannel dataset */
virtual void dscreate( const int rows, const int cols, const int type, String dslabel,
virtual void dscreate( const int rows, const int cols, const int type, const String& dslabel,
const int compresslevel, const int* dims_chunks ) const;
// overload dscreate() #1
virtual void dscreate( const int n_dims, const int* sizes, const int type,
String dslabel ) const;
const String& dslabel ) const;
// overload dscreate() #2
virtual void dscreate( const int n_dims, const int* sizes, const int type,
String dslabel, const int compresslevel ) const;
const String& dslabel, const int compresslevel ) const;
// overload dscreate() #3
virtual void dscreate( const vector<int>& sizes, const int type, String dslabel,
virtual void dscreate( const vector<int>& sizes, const int type, const String& dslabel,
const int compresslevel = H5_NONE, const vector<int>& dims_chunks = vector<int>() ) const;
/* create n-dimensional single or mutichannel dataset */
virtual void dscreate( const int n_dims, const int* sizes, const int type,
String dslabel, const int compresslevel, const int* dims_chunks ) const;
const String& dslabel, const int compresslevel, const int* dims_chunks ) const;
// overload dswrite() #1
virtual void dswrite( InputArray Array, String dslabel ) const;
virtual void dswrite( InputArray Array, const String& dslabel ) const;
// overload dswrite() #2
virtual void dswrite( InputArray Array, String dslabel, const int* dims_offset ) const;
virtual void dswrite( InputArray Array, const String& dslabel, const int* dims_offset ) const;
// overload dswrite() #3
virtual void dswrite( InputArray Array, String dslabel, const vector<int>& dims_offset,
virtual void dswrite( InputArray Array, const String& dslabel, const vector<int>& dims_offset,
const vector<int>& dims_counts = vector<int>() ) const;
/* write into dataset */
virtual void dswrite( InputArray Array, String dslabel,
virtual void dswrite( InputArray Array, const String& dslabel,
const int* dims_offset, const int* dims_counts ) const;
// overload dsinsert() #1
virtual void dsinsert( InputArray Array, String dslabel ) const;
virtual void dsinsert( InputArray Array, const String& dslabel ) const;
// overload dsinsert() #2
virtual void dsinsert( InputArray Array, String dslabel, const int* dims_offset ) const;
virtual void dsinsert( InputArray Array, const String& dslabel, const int* dims_offset ) const;
// overload dsinsert() #3
virtual void dsinsert( InputArray Array, String dslabel,
virtual void dsinsert( InputArray Array, const String& dslabel,
const vector<int>& dims_offset, const vector<int>& dims_counts = vector<int>() ) const;
/* append / merge into dataset */
virtual void dsinsert( InputArray Array, String dslabel,
virtual void dsinsert( InputArray Array, const String& dslabel,
const int* dims_offset = NULL, const int* dims_counts = NULL ) const;
// overload dsread() #1
virtual void dsread( OutputArray Array, String dslabel ) const;
virtual void dsread( OutputArray Array, const String& dslabel ) const;
// overload dsread() #2
virtual void dsread( OutputArray Array, String dslabel, const int* dims_offset ) const;
virtual void dsread( OutputArray Array, const String& dslabel, const int* dims_offset ) const;
// overload dsread() #3
virtual void dsread( OutputArray Array, String dslabel,
virtual void dsread( OutputArray Array, const String& dslabel,
const vector<int>& dims_offset, const vector<int>& dims_counts = vector<int>() ) const;
// read from dataset
virtual void dsread( OutputArray Array, String dslabel,
virtual void dsread( OutputArray Array, const String& dslabel,
const int* dims_offset, const int* dims_counts ) const;
/*
......@@ -156,36 +156,36 @@ public:
*/
// get size of keypoints dataset
virtual int kpgetsize( String kplabel, int dims_flag = H5_GETDIMS ) const;
virtual int kpgetsize( const String& kplabel, int dims_flag = H5_GETDIMS ) const;
// create KeyPoint structure
virtual void kpcreate( const int size, String kplabel,
virtual void kpcreate( const int size, const String& kplabel,
const int compresslevel = H5_NONE, const int chunks = H5_NONE ) const;
// write KeyPoint structures
virtual void kpwrite( const vector<KeyPoint> keypoints, String kplabel,
virtual void kpwrite( const vector<KeyPoint> keypoints, const String& kplabel,
const int offset = H5_NONE, const int counts = H5_NONE ) const;
// append / merge KeyPoint structures
virtual void kpinsert( const vector<KeyPoint> keypoints, String kplabel,
virtual void kpinsert( const vector<KeyPoint> keypoints, const String& kplabel,
const int offset = H5_NONE, const int counts = H5_NONE ) const;
// read KeyPoint structure
virtual void kpread( vector<KeyPoint>& keypoints, String kplabel,
virtual void kpread( vector<KeyPoint>& keypoints, const String& kplabel,
const int offset = H5_NONE, const int counts = H5_NONE ) const;
private:
// store filename
//! store filename
String m_hdf5_filename;
// hdf5 file handler
//! hdf5 file handler
hid_t m_h5_file_id;
// translate cvType -> h5Type
//! translate cvType -> h5Type
inline hid_t GetH5type( int cvType ) const;
// translate h5Type -> cvType
//! translate h5Type -> cvType
inline int GetCVtype( hid_t h5Type ) const;
};
......@@ -247,7 +247,7 @@ inline int HDF5Impl::GetCVtype( hid_t h5Type ) const
return cvType;
}
HDF5Impl::HDF5Impl( String _hdf5_filename )
HDF5Impl::HDF5Impl( const String& _hdf5_filename )
: m_hdf5_filename( _hdf5_filename )
{
// save old
......@@ -260,7 +260,7 @@ HDF5Impl::HDF5Impl( String _hdf5_filename )
// turn off error handling
H5Eset_auto( stackid, NULL, NULL );
// check HDF5 file presence (err supressed)
// check HDF5 file presence (err suppressed)
htri_t check = H5Fis_hdf5( m_hdf5_filename.c_str() );
// restore previous error handler
......@@ -290,7 +290,7 @@ void HDF5Impl::close()
* h5 generic
*/
bool HDF5Impl::hlexists( String label ) const
bool HDF5Impl::hlexists( const String& label ) const
{
bool exists = false;
......@@ -306,7 +306,7 @@ bool HDF5Impl::hlexists( String label ) const
* h5 group
*/
void HDF5Impl::grcreate( String grlabel )
void HDF5Impl::grcreate( const String& grlabel )
{
hid_t gid = H5Gcreate( m_h5_file_id, grlabel.c_str(),
H5P_DEFAULT, H5P_DEFAULT, H5P_DEFAULT);
......@@ -317,7 +317,7 @@ void HDF5Impl::grcreate( String grlabel )
* cv:Mat
*/
vector<int> HDF5Impl::dsgetsize( String dslabel, int dims_flag ) const
vector<int> HDF5Impl::dsgetsize( const String& dslabel, int dims_flag ) const
{
// open dataset
hid_t dsdata = H5Dopen( m_h5_file_id, dslabel.c_str(), H5P_DEFAULT );
......@@ -372,7 +372,7 @@ vector<int> HDF5Impl::dsgetsize( String dslabel, int dims_flag ) const
return SizeVect;
}
int HDF5Impl::dsgettype( String dslabel ) const
int HDF5Impl::dsgettype( const String& dslabel ) const
{
hid_t h5type;
......@@ -408,7 +408,7 @@ int HDF5Impl::dsgettype( String dslabel ) const
// overload
void HDF5Impl::dscreate( const int rows, const int cols, const int type,
String dslabel ) const
const String& dslabel ) const
{
// dataset dims
int dsizes[2] = { rows, cols };
......@@ -419,7 +419,7 @@ void HDF5Impl::dscreate( const int rows, const int cols, const int type,
// overload
void HDF5Impl::dscreate( const int rows, const int cols, const int type,
String dslabel, const int compresslevel ) const
const String& dslabel, const int compresslevel ) const
{
// dataset dims
int dsizes[2] = { rows, cols };
......@@ -430,7 +430,7 @@ void HDF5Impl::dscreate( const int rows, const int cols, const int type,
// overload
void HDF5Impl::dscreate( const int rows, const int cols, const int type,
String dslabel, const int compresslevel,
const String& dslabel, const int compresslevel,
const vector<int>& dims_chunks ) const
{
CV_Assert( dims_chunks.empty() || dims_chunks.size() == 2 );
......@@ -438,7 +438,7 @@ void HDF5Impl::dscreate( const int rows, const int cols, const int type,
}
void HDF5Impl::dscreate( const int rows, const int cols, const int type,
String dslabel, const int compresslevel, const int* dims_chunks ) const
const String& dslabel, const int compresslevel, const int* dims_chunks ) const
{
// dataset dims
int dsizes[2] = { rows, cols };
......@@ -449,21 +449,21 @@ void HDF5Impl::dscreate( const int rows, const int cols, const int type,
// overload
void HDF5Impl::dscreate( const int n_dims, const int* sizes, const int type,
String dslabel ) const
const String& dslabel ) const
{
dscreate( n_dims, sizes, type, dslabel, H5_NONE, NULL );
}
// overload
void HDF5Impl::dscreate( const int n_dims, const int* sizes, const int type,
String dslabel, const int compresslevel ) const
const String& dslabel, const int compresslevel ) const
{
dscreate( n_dims, sizes, type, dslabel, compresslevel, NULL );
}
// overload
void HDF5Impl::dscreate( const vector<int>& sizes, const int type,
String dslabel, const int compresslevel,
const String& dslabel, const int compresslevel,
const vector<int>& dims_chunks ) const
{
CV_Assert( dims_chunks.empty() || dims_chunks.size() == sizes.size() );
......@@ -473,7 +473,7 @@ void HDF5Impl::dscreate( const vector<int>& sizes, const int type,
}
void HDF5Impl::dscreate( const int n_dims, const int* sizes, const int type,
String dslabel, const int compresslevel, const int* dims_chunks ) const
const String& dslabel, const int compresslevel, const int* dims_chunks ) const
{
// compress valid H5_NONE, 0-9
CV_Assert( compresslevel >= H5_NONE && compresslevel <= 9 );
......@@ -552,27 +552,27 @@ void HDF5Impl::dscreate( const int n_dims, const int* sizes, const int type,
}
// overload
void HDF5Impl::dsread( OutputArray Array, String dslabel ) const
void HDF5Impl::dsread( OutputArray Array, const String& dslabel ) const
{
dsread( Array, dslabel, NULL, NULL );
}
// overload
void HDF5Impl::dsread( OutputArray Array, String dslabel,
void HDF5Impl::dsread( OutputArray Array, const String& dslabel,
const int* dims_offset ) const
{
dsread( Array, dslabel, dims_offset, NULL );
}
// overload
void HDF5Impl::dsread( OutputArray Array, String dslabel,
void HDF5Impl::dsread( OutputArray Array, const String& dslabel,
const vector<int>& dims_offset,
const vector<int>& dims_counts ) const
{
dsread( Array, dslabel, &dims_offset[0], &dims_counts[0] );
}
void HDF5Impl::dsread( OutputArray Array, String dslabel,
void HDF5Impl::dsread( OutputArray Array, const String& dslabel,
const int* dims_offset, const int* dims_counts ) const
{
// only Mat support
......@@ -672,25 +672,25 @@ void HDF5Impl::dsread( OutputArray Array, String dslabel,
}
// overload
void HDF5Impl::dswrite( InputArray Array, String dslabel ) const
void HDF5Impl::dswrite( InputArray Array, const String& dslabel ) const
{
dswrite( Array, dslabel, NULL, NULL );
}
// overload
void HDF5Impl::dswrite( InputArray Array, String dslabel,
void HDF5Impl::dswrite( InputArray Array, const String& dslabel,
const int* dims_offset ) const
{
dswrite( Array, dslabel, dims_offset, NULL );
}
// overload
void HDF5Impl::dswrite( InputArray Array, String dslabel,
void HDF5Impl::dswrite( InputArray Array, const String& dslabel,
const vector<int>& dims_offset,
const vector<int>& dims_counts ) const
{
dswrite( Array, dslabel, &dims_offset[0], &dims_counts[0] );
}
void HDF5Impl::dswrite( InputArray Array, String dslabel,
void HDF5Impl::dswrite( InputArray Array, const String& dslabel,
const int* dims_offset, const int* dims_counts ) const
{
// only Mat support
......@@ -715,6 +715,9 @@ void HDF5Impl::dswrite( InputArray Array, String dslabel,
dsdims[d] = matrix.size[d];
}
// FixMe: If one of the groups the dataset belongs to does not exist,
// FixMe: dscreate() will fail!
// FixMe: It should be an error if the specified dataset has not been created instead of trying to create it
// pre-create dataset if needed
if ( hlexists( dslabel ) == false )
dscreate( n_dims, dsizes, matrix.type(), dslabel );
......@@ -771,27 +774,27 @@ void HDF5Impl::dswrite( InputArray Array, String dslabel,
}
// overload
void HDF5Impl::dsinsert( InputArray Array, String dslabel ) const
void HDF5Impl::dsinsert( InputArray Array, const String& dslabel ) const
{
dsinsert( Array, dslabel, NULL, NULL );
}
// overload
void HDF5Impl::dsinsert( InputArray Array, String dslabel,
void HDF5Impl::dsinsert( InputArray Array, const String& dslabel,
const int* dims_offset ) const
{
dsinsert( Array, dslabel, dims_offset, NULL );
}
// overload
void HDF5Impl::dsinsert( InputArray Array, String dslabel,
void HDF5Impl::dsinsert( InputArray Array, const String& dslabel,
const vector<int>& dims_offset,
const vector<int>& dims_counts ) const
{
dsinsert( Array, dslabel, &dims_offset[0], &dims_counts[0] );
}
void HDF5Impl::dsinsert( InputArray Array, String dslabel,
void HDF5Impl::dsinsert( InputArray Array, const String& dslabel,
const int* dims_offset, const int* dims_counts ) const
{
// only Mat support
......@@ -859,7 +862,7 @@ void HDF5Impl::dsinsert( InputArray Array, String dslabel,
// add offset
if ( dims_offset != NULL )
nwdims[d] += dims_offset[d];
// add counts or matrixsize
// add counts or matrix size
if ( dims_counts != NULL )
nwdims[d] += dims_counts[d];
else
......@@ -910,7 +913,7 @@ void HDF5Impl::dsinsert( InputArray Array, String dslabel,
* std::vector<cv::KeyPoint>
*/
int HDF5Impl::kpgetsize( String kplabel, int dims_flag ) const
int HDF5Impl::kpgetsize( const String& kplabel, int dims_flag ) const
{
vector<int> sizes = dsgetsize( kplabel, dims_flag );
......@@ -919,7 +922,7 @@ int HDF5Impl::kpgetsize( String kplabel, int dims_flag ) const
return sizes[0];
}
void HDF5Impl::kpcreate( const int size, String kplabel,
void HDF5Impl::kpcreate( const int size, const String& kplabel,
const int compresslevel, const int chunks ) const
{
// size valid
......@@ -992,7 +995,7 @@ void HDF5Impl::kpcreate( const int size, String kplabel,
H5Sclose( dspace );
}
void HDF5Impl::kpwrite( const vector<KeyPoint> keypoints, String kplabel,
void HDF5Impl::kpwrite( const vector<KeyPoint> keypoints, const String& kplabel,
const int offset, const int counts ) const
{
CV_Assert( keypoints.size() > 0 );
......@@ -1048,7 +1051,7 @@ void HDF5Impl::kpwrite( const vector<KeyPoint> keypoints, String kplabel,
H5Dclose( dsdata );
}
void HDF5Impl::kpinsert( const vector<KeyPoint> keypoints, String kplabel,
void HDF5Impl::kpinsert( const vector<KeyPoint> keypoints, const String& kplabel,
const int offset, const int counts ) const
{
CV_Assert( keypoints.size() > 0 );
......@@ -1132,7 +1135,7 @@ void HDF5Impl::kpinsert( const vector<KeyPoint> keypoints, String kplabel,
H5Dclose( dsdata );
}
void HDF5Impl::kpread( vector<KeyPoint>& keypoints, String kplabel,
void HDF5Impl::kpread( vector<KeyPoint>& keypoints, const String& kplabel,
const int offset, const int counts ) const
{
CV_Assert( keypoints.size() == 0 );
......@@ -1193,7 +1196,7 @@ void HDF5Impl::kpread( vector<KeyPoint>& keypoints, String kplabel,
H5Dclose( dsdata );
}
CV_EXPORTS Ptr<HDF5> open( String HDF5Filename )
CV_EXPORTS Ptr<HDF5> open( const String& HDF5Filename )
{
return makePtr<HDF5Impl>( HDF5Filename );
}
......
modules/hdf/test/test_hdf5.cpp 0 → 100644
View file @ 32aad039
// This file is part of OpenCV project.
// It is subject to the license terms in the LICENSE file found in the top-level directory
// of this distribution and at http://opencv.org/license.html.
/**
* @file test_hdf5.cpp
* @author Fangjun Kuang <csukuangfj dot at gmail dot com>
* @date December 2017
*
*/
#include<stdio.h> // for remove()
#include "test_precomp.hpp"
#include <vector>
using namespace cv;
struct HDF5_Test : public testing::Test
{
virtual void SetUp()
{
m_filename = "test.h5";
// 0 1 2
// 3 4 5
m_single_channel.create(2, 3, CV_32F);
for (size_t i = 0; i < m_single_channel.total(); i++)
{
((float*)m_single_channel.data)[i] = i;
}
// 0 1 2 3 4 5
// 6 7 8 9 10 11
m_two_channels.create(2, 3, CV_32SC2);
for (size_t i = 0; i < m_two_channels.total()*m_two_channels.channels(); i++)
{
((int*)m_two_channels.data)[i] = (int)i;
}
}
//! Remove the hdf5 file
void reset()
{
remove(m_filename.c_str());
}
String m_filename; //!< filename for testing
Ptr<hdf::HDF5> m_hdf_io; //!< HDF5 file pointer
Mat m_single_channel; //!< single channel matrix for test
Mat m_two_channels; //!< two-channel matrix for test
};
TEST_F(HDF5_Test, create_a_single_group)
{
reset();
String group_name = "parent";
m_hdf_io = hdf::open(m_filename);
m_hdf_io->grcreate(group_name);
EXPECT_EQ(m_hdf_io->hlexists(group_name), true);
EXPECT_EQ(m_hdf_io->hlexists("child"), false);
m_hdf_io->close();
}
TEST_F(HDF5_Test, create_a_child_group)
{
reset();
String parent = "parent";
String child = parent + "/child";
m_hdf_io = hdf::open(m_filename);
m_hdf_io->grcreate(parent);
m_hdf_io->grcreate(child);
EXPECT_EQ(m_hdf_io->hlexists(parent), true);
EXPECT_EQ(m_hdf_io->hlexists(child), true);
m_hdf_io->close();
}
TEST_F(HDF5_Test, create_dataset)
{
reset();
String dataset_single_channel = "/single";
String dataset_two_channels = "/dual";
m_hdf_io = hdf::open(m_filename);
m_hdf_io->dscreate(m_single_channel.rows,
m_single_channel.cols,
m_single_channel.type(),
dataset_single_channel);
m_hdf_io->dscreate(m_two_channels.rows,
m_two_channels.cols,
m_two_channels.type(),
dataset_two_channels);
EXPECT_EQ(m_hdf_io->hlexists(dataset_single_channel), true);
EXPECT_EQ(m_hdf_io->hlexists(dataset_two_channels), true);
std::vector<int> dims;
dims = m_hdf_io->dsgetsize(dataset_single_channel, hdf::HDF5::H5_GETDIMS);
EXPECT_EQ(dims.size(), (size_t)2);
EXPECT_EQ(dims[0], m_single_channel.rows);
EXPECT_EQ(dims[1], m_single_channel.cols);
dims = m_hdf_io->dsgetsize(dataset_two_channels, hdf::HDF5::H5_GETDIMS);
EXPECT_EQ(dims.size(), (size_t)2);
EXPECT_EQ(dims[0], m_two_channels.rows);
EXPECT_EQ(dims[1], m_two_channels.cols);
int type;
type = m_hdf_io->dsgettype(dataset_single_channel);
EXPECT_EQ(type, m_single_channel.type());
type = m_hdf_io->dsgettype(dataset_two_channels);
EXPECT_EQ(type, m_two_channels.type());
m_hdf_io->close();
}
TEST_F(HDF5_Test, write_read_dataset_1)
{
reset();
String dataset_single_channel = "/single";
String dataset_two_channels = "/dual";
m_hdf_io = hdf::open(m_filename);
// since the dataset is under the root group, it is created by dswrite() automatically.
m_hdf_io->dswrite(m_single_channel, dataset_single_channel);
m_hdf_io->dswrite(m_two_channels, dataset_two_channels);
EXPECT_EQ(m_hdf_io->hlexists(dataset_single_channel), true);
EXPECT_EQ(m_hdf_io->hlexists(dataset_two_channels), true);
// read single channel matrix
Mat single;
m_hdf_io->dsread(single, dataset_single_channel);
EXPECT_EQ(single.type(), m_single_channel.type());
EXPECT_EQ(single.size(), m_single_channel.size());
EXPECT_NEAR(norm(single-m_single_channel), 0, 1e-10);
// read dual channel matrix
Mat dual;
m_hdf_io->dsread(dual, dataset_two_channels);
EXPECT_EQ(dual.type(), m_two_channels.type());
EXPECT_EQ(dual.size(), m_two_channels.size());
EXPECT_NEAR(norm(dual-m_two_channels), 0, 1e-10);
m_hdf_io->close();
}
TEST_F(HDF5_Test, write_read_dataset_2)
{
reset();
// create the dataset manually if it is not inside
// the root group
String parent = "/parent";
String dataset_single_channel = parent + "/single";
String dataset_two_channels = parent + "/dual";
m_hdf_io = hdf::open(m_filename);
m_hdf_io->grcreate(parent);
EXPECT_EQ(m_hdf_io->hlexists(parent), true);
m_hdf_io->dscreate(m_single_channel.rows,
m_single_channel.cols,
m_single_channel.type(),
dataset_single_channel);
m_hdf_io->dscreate(m_two_channels.rows,
m_two_channels.cols,
m_two_channels.type(),
dataset_two_channels);
EXPECT_EQ(m_hdf_io->hlexists(dataset_single_channel), true);
EXPECT_EQ(m_hdf_io->hlexists(dataset_two_channels), true);
m_hdf_io->dswrite(m_single_channel, dataset_single_channel);
m_hdf_io->dswrite(m_two_channels, dataset_two_channels);
EXPECT_EQ(m_hdf_io->hlexists(dataset_single_channel), true);
EXPECT_EQ(m_hdf_io->hlexists(dataset_two_channels), true);
// read single channel matrix
Mat single;
m_hdf_io->dsread(single, dataset_single_channel);
EXPECT_EQ(single.type(), m_single_channel.type());
EXPECT_EQ(single.size(), m_single_channel.size());
EXPECT_NEAR(norm(single-m_single_channel), 0, 1e-10);
// read dual channel matrix
Mat dual;
m_hdf_io->dsread(dual, dataset_two_channels);
EXPECT_EQ(dual.type(), m_two_channels.type());
EXPECT_EQ(dual.size(), m_two_channels.size());
EXPECT_NEAR(norm(dual-m_two_channels), 0, 1e-10);
m_hdf_io->close();
}
modules/hdf/test/test_main.cpp 0 → 100644
View file @ 32aad039
// This file is part of OpenCV project.
// It is subject to the license terms in the LICENSE file found in the top-level directory
// of this distribution and at http://opencv.org/license.html.
#include "test_precomp.hpp"
CV_TEST_MAIN("cv")
\ No newline at end of file
modules/hdf/test/test_precomp.hpp 0 → 100644
View file @ 32aad039
// This file is part of OpenCV project.
// It is subject to the license terms in the LICENSE file found in the top-level directory
// of this distribution and at http://opencv.org/license.html.
#ifdef __GNUC__
# pragma GCC diagnostic ignored "-Wmissing-declarations"
# if defined __clang__ || defined __APPLE__
# pragma GCC diagnostic ignored "-Wmissing-prototypes"
# pragma GCC diagnostic ignored "-Wextra"
# endif
#endif
#ifndef __OPENCV_TEST_PRECOMP_HPP__
#define __OPENCV_TEST_PRECOMP_HPP__
#include "opencv2/ts.hpp"
#include "opencv2/core.hpp"
#include "opencv2/hdf.hpp"
#endif
modules/hdf/tutorials/create_groups/how_to_create_groups.markdown 0 → 100644
View file @ 32aad039
Creating Groups {#tutorial_hdf_create_groups}
===============================
Goal
----
This tutorial will show you:
- How to create a HDF5 file?
- How to create a group?
- How to check whether a given group exists or not?
- How to create a subgroup?
Source Code
----
The following code creates two groups: `Group1` and `SubGroup1`, where
`SubGroup1` is a child of `Group1`.
You can download the code from [here][1] or find it in the file
`modules/hdf/samples/create_groups.cpp` of the opencv_contrib source code library.
@snippet samples/create_groups.cpp tutorial
Explanation
----
First, we create a HDF5 file
@snippet samples/create_groups.cpp tutorial_create_file
If the given file does not exist, it will be created. Otherwise, it is open for read and write.
Next, we create the group `Group1`
@snippet samples/create_groups.cpp tutorial_create_group
Note that we have to check whether `/Group1` exists or not using
the function `hlexists` before creating it. You can not create
a group with an existing name. Otherwise, an error will occur.
Then, we create the subgroup named `Subgroup1`. In order to
indicate that it is a sub group of `Group1`, we have to
use the group name `/Group1/SubGroup1`:
@snippet samples/create_groups.cpp tutorial_create_subgroup
Note that before creating a subgroup, we have to make sure
that its parent group exists. Otherwise, an error will occur.
In the end, we have to close the file
@snippet samples/create_groups.cpp tutorial_close_file
Result
----
There are many tools that can be used to inspect a given HDF file, such
as HDFView and h5dump. If you are using Ubuntu, you can install
them with the following commands:
@code
sudo apt-get install hdf5-tools hdfview
@endcode
There are also binaries available from the The HDF Group official website <https://support.hdfgroup.org/HDF5/Tutor/tools.html>.
The following figure shows the result visualized with the tool HDFView:
![Figure 1: Results of creating groups and subgroups](pics/create_groups.png)
The output for `h5dump` is:
@code
$ h5dump mytest.h5
HDF5 "mytest.h5" {
GROUP "/" {
GROUP "Group1" {
GROUP "SubGroup1" {
}
}
}
}
@endcode
[1]: https://github.com/opencv/opencv_contrib/tree/master/modules/hdf/samples/create_groups.cpp
modules/hdf/tutorials/create_read_write_dataset/create_read_write_dataset.markdown 0 → 100644
View file @ 32aad039
Creating, Writing and Reading Datasets {#tutorial_hdf_create_read_write_datasets}
===============================
Goal
----
This tutorial shows you:
- How to create a dataset?
- How to write a `cv::Mat` to a dataset?
- How to read a `cv::Mat` from a dataset?
@note Currently, it supports only reading and writing `cv::Mat` and the matrix should be continuous
in memory. Supports for other data types have not been implemented yet.
Source Code
----
The following code demonstrates writing a single channel
matrix and a two-channel matrix to datasets and then reading them
back.
You can download the code from [here][1] or find it in the file
`modules/hdf/samples/create_read_write_datasets.cpp` of the opencv_contrib source code library.
@snippet samples/create_read_write_datasets.cpp tutorial
Explanation
----
The first step for creating a dataset is to open the file
@snippet samples/create_read_write_datasets.cpp tutorial_open_file
For the function `write_root_group_single_channel()`, since
the dataset name is `/single`, which is inside the root group, we can use
@snippet samples/create_read_write_datasets.cpp tutorial_write_root_single_channel
to write the data directly to the dataset without the need of creating
it beforehand. Because it is created inside `HDF5::dswrite()`
automatically.
@warning This applies only to datasets that reside inside the root group.
Of course, we can create the dataset by ourselves:
@snippet samples/create_read_write_datasets.cpp tutorial_create_dataset
To read data from a dataset, we use
@snippet samples/create_read_write_datasets.cpp tutorial_read_dataset
by specifying the name of the dataset.
We can check that the data read out is exactly the data written before by using
@snippet samples/create_read_write_datasets.cpp tutorial_check_result
Results
----
Figure 1 shows the result visualized using the tool HDFView for the file
`root_group_sinle_channel`. The results
of matrices for datasets that are not the direct children of the root group
are given in Figure 2 and Figure 3, respectively.
![Figure 1: Result for writing a single channel matrix to a dataset inside the root group](pics/root_group_single_channel.png)
![Figure 2: Result for writing a single channel matrix to a dataset not in the root group](pics/single_channel.png)
![Figure 3: Result for writing a two-channel matrix to a dataset not in the root group](pics/two_channels.png)
[1]: https://github.com/opencv/opencv_contrib/tree/master/modules/hdf/samples/create_read_write_datasets.cpp
modules/hdf/tutorials/table_of_content_hdf.markdown 0 → 100644
View file @ 32aad039
The Hierarchical Data Format (hdf) I/O {#tutorial_table_of_content_hdf}
=====================================
Here you will know how to read and write a HDF5 file using OpenCV.
Currently, only `cv::Mat` is supported.
Note that the HDF5 library has to be installed in your system
to use this module.
- @subpage tutorial_hdf_create_groups
*Compatibility:* \> OpenCV 3.0
*Author:* Fangjun Kuang
You will learn how to create groups and subgroups.
- @subpage tutorial_hdf_create_read_write_datasets
*Compatibility:* \> OpenCV 3.0
*Author:* Fangjun Kuang
You will learn how to create, read and write datasets.
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment