You are here: Start » Function Reference » Computer Vision » Image Segmentation » SegmentImage_Gray_Linear

SegmentImage_Gray_Linear

Header:	AVL.h
Namespace:	avl
Module:	FoundationPro

Segments an image into blobs examining differences between pixels values.

Applications: Detection of objects of undefined shape, but characterized by uniform brightness and good contrast to the background.

Syntax

void avl::SegmentImage_Gray_Linear
(
	const avl::Image& inImage,
	atl::Optional<const avl::Region&> inRoi,
	avl::BlobMergingMethod::Type inMergingMethod,
	avl::RegionConnectivity::Type inConnectivity,
	int inMaxDifference,
	int inHysteresis,
	int inPassCount,
	bool inDirectional,
	int inMinArea,
	atl::Optional<int> inMaxArea,
	atl::Array<avl::Region>& outBlobs
)

Parameters

Name	Type	Range	Default	Description
inImage	const Image&			Input image
inRoi	Optional<const Region&>		NIL	Range of pixels to be processed
inMergingMethod	BlobMergingMethod::Type		Neighbor	Defines a criterion by which two pixels can be merged into one blob
inConnectivity	RegionConnectivity::Type			Defines if pixels can be merged only horizontally and vertically, or also diagonally
inMaxDifference	int	0 -	5	Maximal tonal difference that allows two pixels to be merged into one blob
inHysteresis	int			Total change of MaxDifference during all image passes
inPassCount	int	1 - 5		Number of iterations in which the entire image is scanned for pixels that can be merged
inDirectional	bool			When 'True' the filter can segment results of GradientDirAndPresenceImage
inMinArea	int	0 -	20	Minimal area of a blob that may be accepted
inMaxArea	Optional<int>		NIL	Minimal area of a blob that may be accepted
outBlobs	Array<Region>&

Requirements

For input inImage only pixel formats are supported: 1⨯uint8, 1⨯int8, 1⨯uint16, 1⨯int16, 1⨯int32, 1⨯real, 3⨯uint8, 3⨯int8, 3⨯uint16, 3⨯int16, 3⨯int32, 3⨯real.

Read more about pixel formats in Image documentation.

Description

The filter segments the inImage image into blobs of pixels which gray values do not differ too much. The classification is different depending on the inMergingMethod input:

Neighbor – two adjacent pixels are considered to belong to the same blob when their values differ by at most inMaxDifference
NeighborVerified – as above, but an additional verification is performed to avoid undersegmentation
Mean – pixel is considered to belong to an adjacent blob when its value differs by at most inMaxDifference from the mean value of this blob's pixels
MeanSorted – not implemented in this filter

Examples

SegmentImage_Gray_Linear performed on the sample image with inMaxDifference = 55, inMergingMethod = MeanLinear and inMinArea = 50.

Errors

List of possible exceptions:

Error type	Description
DomainError	Directional method only supports single-channel images.
DomainError	Directional method only supports UInt8 images.
DomainError	Two- and four-channel images are not supported in SegmentImage_Gray.
DomainError	Not supported inImage pixel format in SegmentImage_Gray_Linear.