DIS optical flow algorithm. More...

#include <opencv2/video/tracking.hpp>

Inheritance diagram for cv::DISOpticalFlow:

Public Types
enum	{ PRESET_ULTRAFAST = 0 , PRESET_FAST = 1 , PRESET_MEDIUM = 2 }

Public Member Functions
virtual int	getFinestScale () const =0
	Finest level of the Gaussian pyramid on which the flow is computed (zero level corresponds to the original image resolution). The final flow is obtained by bilinear upscaling.

virtual int	getGradientDescentIterations () const =0
	Maximum number of gradient descent iterations in the patch inverse search stage. Higher values may improve quality in some cases.

virtual int	getPatchSize () const =0
	Size of an image patch for matching (in pixels). Normally, default 8x8 patches work well enough in most cases.

virtual int	getPatchStride () const =0
	Stride between neighbor patches. Must be less than patch size. Lower values correspond to higher flow quality.

virtual bool	getUseMeanNormalization () const =0
	Whether to use mean-normalization of patches when computing patch distance. It is turned on by default as it typically provides a noticeable quality boost because of increased robustness to illumination variations. Turn it off if you are certain that your sequence doesn't contain any changes in illumination.

virtual bool	getUseSpatialPropagation () const =0
	Whether to use spatial propagation of good optical flow vectors. This option is turned on by default, as it tends to work better on average and can sometimes help recover from major errors introduced by the coarse-to-fine scheme employed by the DIS optical flow algorithm. Turning this option off can make the output flow field a bit smoother, however.

virtual float	getVariationalRefinementAlpha () const =0
	Weight of the smoothness term.

virtual float	getVariationalRefinementDelta () const =0
	Weight of the color constancy term.

virtual float	getVariationalRefinementGamma () const =0
	Weight of the gradient constancy term.

virtual int	getVariationalRefinementIterations () const =0
	Number of fixed point iterations of variational refinement per scale. Set to zero to disable variational refinement completely. Higher values will typically result in more smooth and high-quality flow.

virtual void	setFinestScale (int val)=0
	Finest level of the Gaussian pyramid on which the flow is computed (zero level corresponds to the original image resolution). The final flow is obtained by bilinear upscaling.

virtual void	setGradientDescentIterations (int val)=0
	Maximum number of gradient descent iterations in the patch inverse search stage. Higher values may improve quality in some cases.

virtual void	setPatchSize (int val)=0
	Size of an image patch for matching (in pixels). Normally, default 8x8 patches work well enough in most cases.

virtual void	setPatchStride (int val)=0
	Stride between neighbor patches. Must be less than patch size. Lower values correspond to higher flow quality.

virtual void	setUseMeanNormalization (bool val)=0
	Whether to use mean-normalization of patches when computing patch distance. It is turned on by default as it typically provides a noticeable quality boost because of increased robustness to illumination variations. Turn it off if you are certain that your sequence doesn't contain any changes in illumination.

virtual void	setUseSpatialPropagation (bool val)=0
	Whether to use spatial propagation of good optical flow vectors. This option is turned on by default, as it tends to work better on average and can sometimes help recover from major errors introduced by the coarse-to-fine scheme employed by the DIS optical flow algorithm. Turning this option off can make the output flow field a bit smoother, however.

virtual void	setVariationalRefinementAlpha (float val)=0
	Weight of the smoothness term.

virtual void	setVariationalRefinementDelta (float val)=0
	Weight of the color constancy term.

virtual void	setVariationalRefinementGamma (float val)=0
	Weight of the gradient constancy term.

virtual void	setVariationalRefinementIterations (int val)=0
	Maximum number of gradient descent iterations in the patch inverse search stage. Higher values may improve quality in some cases.

Public Member Functions inherited from cv::DenseOpticalFlow
virtual void	calc (InputArray I0, InputArray I1, InputOutputArray flow)=0
	Calculates an optical flow.

virtual void	collectGarbage ()=0
	Releases all inner buffers.

Public Member Functions inherited from cv::Algorithm
	Algorithm ()

virtual	~Algorithm ()

virtual void	clear ()
	Clears the algorithm state.

virtual bool	empty () const
	Returns true if the Algorithm is empty (e.g. in the very beginning or after unsuccessful read.

virtual String	getDefaultName () const

virtual void	read (const FileNode &fn)
	Reads algorithm parameters from a file storage.

virtual void	save (const String &filename) const

void	write (const Ptr< FileStorage > &fs, const String &name=String()) const

virtual void	write (FileStorage &fs) const
	Stores algorithm parameters in a file storage.

void	write (FileStorage &fs, const String &name) const

Static Public Member Functions
static Ptr< DISOpticalFlow >	create (int preset=DISOpticalFlow::PRESET_FAST)
	Creates an instance of DISOpticalFlow.

Static Public Member Functions inherited from cv::Algorithm
template<typename _Tp >
static Ptr< _Tp >	load (const String &filename, const String &objname=String())
	Loads algorithm from the file.

template<typename _Tp >
static Ptr< _Tp >	loadFromString (const String &strModel, const String &objname=String())
	Loads algorithm from a String.

template<typename _Tp >
static Ptr< _Tp >	read (const FileNode &fn)
	Reads algorithm from the file node.

Additional Inherited Members
Protected Member Functions inherited from cv::Algorithm
void	writeFormat (FileStorage &fs) const

Detailed Description

DIS optical flow algorithm.

This class implements the Dense Inverse Search (DIS) optical flow algorithm. More details about the algorithm can be found at [Kroeger2016] . Includes three presets with preselected parameters to provide reasonable trade-off between speed and quality. However, even the slowest preset is still relatively fast, use DeepFlow if you need better quality and don't care about speed.

This implementation includes several additional features compared to the algorithm described in the paper, including spatial propagation of flow vectors (getUseSpatialPropagation), as well as an option to utilize an initial flow approximation passed to calc (which is, essentially, temporal propagation, if the previous frame's flow field is passed).

Member Enumeration Documentation

◆ anonymous enum

anonymous enum

Enumerator
PRESET_ULTRAFAST
PRESET_FAST
PRESET_MEDIUM

Member Function Documentation

◆ create()

static Ptr< DISOpticalFlow > cv::DISOpticalFlow::create ( int preset = DISOpticalFlow::PRESET_FAST )

static

Creates an instance of DISOpticalFlow.

Parameters

preset one of PRESET_ULTRAFAST, PRESET_FAST and PRESET_MEDIUM

◆ getFinestScale()

virtual int cv::DISOpticalFlow::getFinestScale ( ) const

pure virtual

Finest level of the Gaussian pyramid on which the flow is computed (zero level corresponds to the original image resolution). The final flow is obtained by bilinear upscaling.

See also: setFinestScale

◆ getGradientDescentIterations()

virtual int cv::DISOpticalFlow::getGradientDescentIterations ( ) const

pure virtual

Maximum number of gradient descent iterations in the patch inverse search stage. Higher values may improve quality in some cases.

See also: setGradientDescentIterations

◆ getPatchSize()

virtual int cv::DISOpticalFlow::getPatchSize ( ) const

pure virtual

Size of an image patch for matching (in pixels). Normally, default 8x8 patches work well enough in most cases.

See also: setPatchSize

◆ getPatchStride()

virtual int cv::DISOpticalFlow::getPatchStride ( ) const

pure virtual

Stride between neighbor patches. Must be less than patch size. Lower values correspond to higher flow quality.

See also: setPatchStride

◆ getUseMeanNormalization()

virtual bool cv::DISOpticalFlow::getUseMeanNormalization ( ) const

pure virtual

Whether to use mean-normalization of patches when computing patch distance. It is turned on by default as it typically provides a noticeable quality boost because of increased robustness to illumination variations. Turn it off if you are certain that your sequence doesn't contain any changes in illumination.

See also: setUseMeanNormalization

◆ getUseSpatialPropagation()

virtual bool cv::DISOpticalFlow::getUseSpatialPropagation ( ) const

pure virtual

Whether to use spatial propagation of good optical flow vectors. This option is turned on by default, as it tends to work better on average and can sometimes help recover from major errors introduced by the coarse-to-fine scheme employed by the DIS optical flow algorithm. Turning this option off can make the output flow field a bit smoother, however.

See also: setUseSpatialPropagation

◆ getVariationalRefinementAlpha()

virtual float cv::DISOpticalFlow::getVariationalRefinementAlpha ( ) const

pure virtual

Weight of the smoothness term.

See also: setVariationalRefinementAlpha

◆ getVariationalRefinementDelta()

virtual float cv::DISOpticalFlow::getVariationalRefinementDelta ( ) const

pure virtual

Weight of the color constancy term.

See also: setVariationalRefinementDelta

◆ getVariationalRefinementGamma()

virtual float cv::DISOpticalFlow::getVariationalRefinementGamma ( ) const

pure virtual

Weight of the gradient constancy term.

See also: setVariationalRefinementGamma

◆ getVariationalRefinementIterations()

virtual int cv::DISOpticalFlow::getVariationalRefinementIterations ( ) const

pure virtual

Number of fixed point iterations of variational refinement per scale. Set to zero to disable variational refinement completely. Higher values will typically result in more smooth and high-quality flow.

See also: setGradientDescentIterations

◆ setFinestScale()

virtual void cv::DISOpticalFlow::setFinestScale ( int val )

pure virtual

Finest level of the Gaussian pyramid on which the flow is computed (zero level corresponds to the original image resolution). The final flow is obtained by bilinear upscaling.

See also: getFinestScale

◆ setGradientDescentIterations()

virtual void cv::DISOpticalFlow::setGradientDescentIterations ( int val )

pure virtual

Maximum number of gradient descent iterations in the patch inverse search stage. Higher values may improve quality in some cases.

See also: getGradientDescentIterations

◆ setPatchSize()

virtual void cv::DISOpticalFlow::setPatchSize ( int val )

pure virtual

Size of an image patch for matching (in pixels). Normally, default 8x8 patches work well enough in most cases.

See also: getPatchSize

◆ setPatchStride()

virtual void cv::DISOpticalFlow::setPatchStride ( int val )

pure virtual

Stride between neighbor patches. Must be less than patch size. Lower values correspond to higher flow quality.

See also: getPatchStride

◆ setUseMeanNormalization()

virtual void cv::DISOpticalFlow::setUseMeanNormalization ( bool val )

pure virtual

Whether to use mean-normalization of patches when computing patch distance. It is turned on by default as it typically provides a noticeable quality boost because of increased robustness to illumination variations. Turn it off if you are certain that your sequence doesn't contain any changes in illumination.

See also: getUseMeanNormalization

◆ setUseSpatialPropagation()

virtual void cv::DISOpticalFlow::setUseSpatialPropagation ( bool val )

pure virtual

Whether to use spatial propagation of good optical flow vectors. This option is turned on by default, as it tends to work better on average and can sometimes help recover from major errors introduced by the coarse-to-fine scheme employed by the DIS optical flow algorithm. Turning this option off can make the output flow field a bit smoother, however.

See also: getUseSpatialPropagation

◆ setVariationalRefinementAlpha()

virtual void cv::DISOpticalFlow::setVariationalRefinementAlpha ( float val )

pure virtual

Weight of the smoothness term.

See also: getVariationalRefinementAlpha

◆ setVariationalRefinementDelta()

virtual void cv::DISOpticalFlow::setVariationalRefinementDelta ( float val )

pure virtual

Weight of the color constancy term.

See also: getVariationalRefinementDelta

◆ setVariationalRefinementGamma()

virtual void cv::DISOpticalFlow::setVariationalRefinementGamma ( float val )

pure virtual

Weight of the gradient constancy term.

See also: getVariationalRefinementGamma

◆ setVariationalRefinementIterations()

virtual void cv::DISOpticalFlow::setVariationalRefinementIterations ( int val )

pure virtual

Maximum number of gradient descent iterations in the patch inverse search stage. Higher values may improve quality in some cases.

See also: getGradientDescentIterations

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

opencv2/video/tracking.hpp

Public Types

Public Member Functions

Static Public Member Functions

Additional Inherited Members

Detailed Description

Member Enumeration Documentation

◆ anonymous enum

Member Function Documentation

◆ create()

◆ getFinestScale()

◆ getGradientDescentIterations()

◆ getPatchSize()

◆ getPatchStride()

◆ getUseMeanNormalization()

◆ getUseSpatialPropagation()

◆ getVariationalRefinementAlpha()

◆ getVariationalRefinementDelta()

◆ getVariationalRefinementGamma()

◆ getVariationalRefinementIterations()

◆ setFinestScale()

◆ setGradientDescentIterations()

◆ setPatchSize()

◆ setPatchStride()

◆ setUseMeanNormalization()

◆ setUseSpatialPropagation()

◆ setVariationalRefinementAlpha()

◆ setVariationalRefinementDelta()

◆ setVariationalRefinementGamma()

◆ setVariationalRefinementIterations()