eLynx::Image::IImageRestoration Class Reference

#include <IImageRestoration.h>

Inheritance diagram for eLynx::Image::IImageRestoration:

List of all members.

Public Member Functions
virtual	~IImageRestoration ()
virtual bool	FastInpaint (AbstractImage &ioImage, const ImageLub &iBinaryMask, bool iUseRegion, uint32 iChannelMask, ProgressNotifier &iNotifier) const =0
virtual bool	FastMarchingInpaint (AbstractImage &ioImage, const ImageLub &iBinaryMask, uint32 iSize, bool iUseRegion, uint32 iChannelMask, ProgressNotifier &iNotifier) const =0
virtual bool	ExemplarBasedInpaint (AbstractImage &ioImage, const ImageLub &iBinaryMask, const Math::AOBBox2i &iSearchArea, uint32 iPatchSize, bool iUseRegion, uint32 iChannelMask, ProgressNotifier &iNotifier) const =0

---

Detailed Description

Interface defining various restoration methods.

Definition at line 71 of file IImageRestoration.h.

---

Constructor & Destructor Documentation

virtual eLynx::Image::IImageRestoration::~IImageRestoration

(

)

[virtual]

---

Member Function Documentation

virtual bool eLynx::Image::IImageRestoration::FastInpaint	(	AbstractImage &	ioImage,
		const ImageLub &	iBinaryMask,
		bool	iUseRegion,
		uint32	iChannelMask,
		ProgressNotifier &	iNotifier
	)			const [pure virtual]

Implementation based on Fast Digital Image Inpainting article.

[1]	Fast Digital Image Inpainting September 3, 2001 by Manuel M. Oliveira, Brian Bowen, Richard McKenna and Yu-Sung Chang Department of Computer Science State University of New York at Stony Brook

Very simple inpainting algorithm based on the convolution of the small image region to be restored which is equivalent to the isotropic diffusion. The convolution kernel is a weighted average 3x3 kernel with a zero weight at the center.

Parameters:

	ioImage	image to be restored.
	iBinaryMask	image bitmap. The pixel's value is either IPT_UNKNOWN for pixels to be inpainted or IPT_KNOWN for known pixels.
	iUseRegion	flag to use optimizations to inpaint only localized regions. Caution: Some images produce better results without this optimization.
	iChannelMask	selected channels to be processed. Default is all.
	iNotifier	a notifier for progression, log and cancel.

Returns:

Method running status.

Implemented in eLynx::Image::ImageRestorationImpl< Pixel >.

virtual bool eLynx::Image::IImageRestoration::FastMarchingInpaint	(	AbstractImage &	ioImage,
		const ImageLub &	iBinaryMask,
		uint32	iSize,
		bool	iUseRegion,
		uint32	iChannelMask,
		ProgressNotifier &	iNotifier
	)			const [pure virtual]

Implementation based on An Image Inpainting Technique Based on the Fast Marching Method article.

[1]	An Image Inpainting Technique Based on the Fast Marching Method May 21, 2003 by Alexandru Telea Eindhoven University of Technology

Simple inpainting algorithm based on the Fast Marching Method (FMM). Pixels are inpainted in the order based on the distance to the know area.

Parameters:

	ioImage	image to be restored.
	iBinaryMask	image bitmap. The pixel's value is either IPT_UNKNOWN for pixels to be inpainted or IPT_KNOWN for known pixels .
	iSize	size of the pixel's neighborhood. Min value is 3.
	iUseRegion	flag to use optimizations to inpaint only localized regions. Caution: Some images produce better results without this optimization.
	iChannelMask	selected channels to be processed. Default is all.
	iNotifier	a notifier for progression, log and cancel.

Returns:

Method running status.

Implemented in eLynx::Image::ImageRestorationImpl< Pixel >.

virtual bool eLynx::Image::IImageRestoration::ExemplarBasedInpaint	(	AbstractImage &	ioImage,
		const ImageLub &	iBinaryMask,
		const Math::AOBBox2i &	iSearchArea,
		uint32	iPatchSize,
		bool	iUseRegion,
		uint32	iChannelMask,
		ProgressNotifier &	iNotifier
	)			const [pure virtual]

Implementation based on Region Filling and Object Removal by Exemplar-Based Image Inpainting article.
Inpainting algorithm to remove int64 objects from images. The algorithm propagets texture and structure into the inpainting region by filling best-fit patches in the order based on the confidence level.

[1]	Region Filling and Object Removal by Exemplar-Based Image Inpainting SEP 2004 by A. Criminisi*, P. Pérez and K. Toyama Microsoft Research, Cambridge (UK) and Redmond (US) IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 13, NO. 9

Parameters:

	ioImage	image to be restored.
	iBinaryMask	image bitmap. The pixel's value is either IPT_UNKNOWN for pixels to be inpainted or IPT_KNOWN for known pixels.
	iSearchArea	bounding box to restrict the image area which would be searched for the patches to fill.
	iPatchSize	size of the patch. Min value is 3.
	iUseRegion	flag to use optimizations to inpaint only localized regions. Caution: Some images produce better results without this optimization.
	iChannelMask	selected channels to be processed. Default is all.
	iNotifier	a notifier for progression, log and cancel.

Returns:

Method running status.

Implemented in eLynx::Image::ImageRestorationImpl< Pixel >.

---

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

• IImageRestoration.h