w2qcref::CombineSmartMean Class Reference

Inheritance diagram for w2qcref::CombineSmartMean:

List of all members.

---

Detailed Description

postprocesses the precip confidence field by computing a local mean, but taking into account the distribution of high confidences in either direction.

Author:

Lakshman

Version:

Id

w2img_Postprocessor.h,v 1.9 2009/06/02 20:18:01 lakshman Exp

Public Member Functions
	CombineSmartMean (float not_precip=0.25, float def_precip=0.75)
	Pass in thresholds for when we should consider the probabilites "definite".
virtual	~CombineSmartMean ()
virtual void	doPostprocess (const code::Image< float > &expanded_input, code::PolarGrid &output, const code::Image< float > &expanded_refmax, int half_size, int full_size, float edge_threshold)
	sub-classes over-ride this method.
void	postprocess (code::PolarGrid &grid, code::Image< float > &refmax)
	postprocesses the input field, replacing its data values.
void	setHalfSize (int hsize)
	change default from 2
void	setEdgeThreshold (float edge)
	change default threshold for precip from 0.
void	useBlobInformation (const code::Image< float > &refmax, code::PolarGrid &grid, const code::Image< int > &labels)
	Use the label information about regions.
virtual void	clear ()
Static Public Member Functions
static void	setBoostingDirectory (const std::string &dir)
	During NN training, polar grids with boosting information will be written out to this directory.
Protected Member Functions
void	writeBoostingProduct (const code::PolarGrid &example, const code::Image< int > &labels, const std::vector< bool > &marginal)

---

Constructor & Destructor Documentation

w2qcref::CombineSmartMean::CombineSmartMean	(	float	not_precip = 0.25,
		float	def_precip = 0.75
	)			[inline]

Pass in thresholds for when we should consider the probabilites "definite".

virtual w2qcref::CombineSmartMean::~CombineSmartMean

(

)

[inline, virtual]

---

Member Function Documentation

virtual void w2qcref::Postprocessor::clear

(

)

[virtual, inherited]

virtual void w2qcref::CombineSmartMean::doPostprocess	(	const code::Image< float > &	expanded_input,
		code::PolarGrid &	output,
		const code::Image< float > &	expanded_refmax,
		int	half_size,
		int	full_size,
		float	edge_threshold
	)			[virtual]

sub-classes over-ride this method.

The expanded_input has wrapped radials.

Implements w2qcref::Postprocessor.

void w2qcref::Postprocessor::postprocess	(	code::PolarGrid &	grid,
		code::Image< float > &	refmax
	)			[inherited]

postprocesses the input field, replacing its data values.

Can also correct the refmax for atroicious values that should not be added back by connectivity-based tests.

static void w2qcref::Postprocessor::setBoostingDirectory

(

const std::string &

dir

)

[inline, static, inherited]

During NN training, polar grids with boosting information will be written out to this directory.

If this method is not called, no boosted products will be written out (assumed to be in runtime mode).

void w2qcref::Postprocessor::setEdgeThreshold

(

float

edge

)

[inline, inherited]

change default threshold for precip from 0.

This value is used to decide edges etc.

void w2qcref::Postprocessor::setHalfSize

(

int

hsize

)

[inline, inherited]

change default from 2

void w2qcref::Postprocessor::useBlobInformation	(	const code::Image< float > &	refmax,
		code::PolarGrid &	grid,
		const code::Image< int > &	labels
	)			[inherited]

Use the label information about regions.

void w2qcref::Postprocessor::writeBoostingProduct	(	const code::PolarGrid &	example,
		const code::Image< int > &	labels,
		const std::vector< bool > &	marginal
	)			[protected, inherited]

---