ILCrossMarkDetector Interface Reference

This interface provide functionalities of sub-pixel cross mark detection. More...

Inheritance diagram for ILCrossMarkDetector:

Public Member Functions
LPVErrorCode	Detect (ILImage *img, ILRegion *region, ILCrossMarkResults **results)
void	SetMarkRange (int grayLB, int grayUB)
void	SetSizeRange (int lb, int ub)
Public Member Functions inherited from ILObject
ILObject *	Copy ()
LPVErrorCode	Load (LString filename)
void	Reset ()
LPVErrorCode	Save (LString filename)
BOOL	Valid ()

Properties
double	FitDistThreshold [get, set]
	The distance threshold is used for the fitting of the cross lines, as the maximum acceptable distance between the fitting results and the valid fitting points. Those exceed the limitation will be treated as outliers and then excluded from the fitting. By default, it's set to 0, means we'll decide a proper threshold on the fly during detection, according the cross mark's size and width.
int	KernelSize [get, set]
	The kernel size used in edge filtering, should be an odd number which is ≥ 3, say 3, 5, 7, etc.
int	MarkRangeLB [get]
	The lower bound of the mark's grayscale value range when custom thresholding is used.
int	MarkRangeUB [get]
	The upper bound of the mark's grayscale value range when custom thresholding is used.
int	MaxCount [get, set]
	The maximum of cross marks to find. If there are more than the setting counts of cross marks, sort the candidates then pick the tops.
int	NoiseThreshold [get, set]
	The noise threshold is used to filter false detections, which is the required minimum luminance difference between the cross mark and the background. It's in 1 ~ 100. By default, it's set to 20.
LPVPolarity	Polarity [get, set]
	The detection polarity, either object or edge polarity is acceptable. Edge polarities are recommended to use when the cross mark's edge is clear and stable. . More...
int	SizeRangeLB [get]
	The lower bound of the mark's size range.
int	SizeRangeUB [get]
	The upper bound of the mark's size range.
BOOL	UseAutoThreshold [get, set]
	Whether to use auto-thresholding for candidate position extraction. By default, it's true. Local-adaptive thresholding is adopted when there's available mark size range, otherwise global-adaptive thresholding is used.
double	Width [get, set]
	The width of the cross mark's two cross lines. By default, it's set to 0, means we'll guess a proper width on the fly during detection.

Detailed Description

This interface provide functionalities of sub-pixel cross mark detection.

Cross mark detector is designed to locate a single or multiple cross mark in given image.
The algorithm firstly extracts the candidate positions according the grayscale difference or edges around the mark, then tries to locate the two crossed midlines of the cross at each candidate position to obtain the intersection point as mark center. For multi-results scenarios, we evaluate each result based on its edge lines' fitting quality and parallelism, as well as the symmetry of the pattern, then the top scores are reported as final results.
The algorithm is real-time and has accuracy of sub-pixel.

To use this interface, you should create a LCrossMarkDetector object.

Example Code

C++

ILCrossMarkDetectorPtr cmDetector = LCrossMarkDetector::Create();
cmDetector->Polarity = LPVPolarity::LPVBlackOnWhite;
cmDetector->Width = 20;
cmDetector->SetSizeRange(100, 150);
cmDetector->KernelSize = 3;
ILCrossMarkResultsPtr cmResults;
LPVErrorCode err = cmDetector->Detect(img, roi, &cmResults);
if (err != LPVErrorCode.LPVNoError) return;

C#

LCrossMarkDetector cmDetector = new LCrossMarkDetector();
cmDetector.Polarity = LPVPolarity.LPVBlackOnWhite;
cmDetector.Width = 20;
cmDetector.SetSizeRange(100, 150);
cmDetector.KernelSize = 3;
ILCrossMarkResultsPtr cmResults;
LPVErrorCode err = cmDetector.Detect(img, roi, out cmResults);
if (err != LPVErrorCode.LPVNoError) return;

COM

ILCrossMarkDetectorPtr cmDetector = LCrossMarkDetector::Create();
// or, ILCrossMarkDetectorPtr cmDetector(__uuidof(LCrossMarkDetector)), for absence of built-in Create functions (before 2.2.0)
cmDetector->Polarity = LPVPolarity::LPVBlackOnWhite;
cmDetector->Width = 20;
cmDetector->SetSizeRange(100, 150);
cmDetector->KernelSize = 3;
ILCrossMarkResultsPtr cmResults;
LPVErrorCode err = cmDetector->Detect(img, roi, &cmResults);
if (err != LPVErrorCode.LPVNoError) return;

Member Function Documentation

Detect()

LPVErrorCode Detect	(	ILImage *	img,
		ILRegion *	region,
		ILCrossMarkResults **	results
	)

Detect cross marks using the cross mark detector.

Parameters

[in]	img	The input image.
[in]	region	The input region.
[out]	results	Return the cross mark results.

Return values

error

Return error code if anything is wrong.

SetMarkRange()

void SetMarkRange	(	int	grayLB,
		int	grayUB
	)

Set the cross mark's grayscale range and use the custom thresholding for candidate position extraction. UseAutoThreshold will be set to false then.

Parameters

[in]	grayLB	The lower bound of the grayscale value range.
[in]	grayUB	The upper bound of the grayscale value range.

See also

MarkRangeLB, MarkRangeUB, UseAutoThreshold

SetSizeRange()

void SetSizeRange	(	int	lb,
		int	ub
	)

Set the cross mark's size range. By default, it's set to an empty (0, 0) range, means we'll guess a proper size on the fly during detection.

Parameters

[in]	lb	The lower bound of the size range.
[in]	ub	The upper bound of the size range.

See also

Width

Property Documentation

Polarity

LPVPolarity Polarity

getset

The detection polarity, either object or edge polarity is acceptable.
Edge polarities are recommended to use when the cross mark's edge is clear and stable.
.

Otherwise, when the mark object is small or blurry or edge transition changes from time to time, use object polarity instead.

See also

KernelSize, UseAutoThreshold