ILOCR Interface Reference

This interface provide functionalities of character segmentation, recognition and verification. More...

Inheritance diagram for ILOCR:

Public Member Functions
LPVErrorCode	Detect (ILImage *img, ILRegion *region, ILOCRResult **results)
LPVErrorCode	DetectAndVerify (ILImage *img, ILRegion *region, ILOCRResult **results)
LPVErrorCode	Segment (ILImage *img, ILRegion *region, ILOCRResult **results)
void	SetCharSize (int w, int h, double wTolerance, double hTolerance)
void	SetCharSkew (double skewLB, double skewUB)
Public Member Functions inherited from ILObject
ILObject *	Copy ()
LPVErrorCode	Load (LString filename)
void	Reset ()
LPVErrorCode	Save (LString filename)
BOOL	Valid ()

Properties
double	AspectRatioChange [get, set]
	The resize ratio compared with the standard aspect ratio in the font database, in range 0.5 ~ 1.5. By default, it's 1 means no change.
BOOL	AutoDotPrint [get, set]
	Whether to turn on automatic dot-print detection. By default, it's off.
int	CharHeightLB [get, set]
	The lower bound of the character's estimated height, at least 1 pixel.
int	CharHeightUB [get, set]
	The upper bound of the character's estimated height, should be larger than or equal to the lower bound.
double	CharSkewLB [get, set]
	The lower bound of the italic character's skew int angle, in -60 ~ 60 degree.
double	CharSkewUB [get, set]
	The upper bound of the italic character's skew int angle, should be in -60 ~ 60 degree and larger than or equal to the lower bound.
int	CharWidthLB [get, set]
	The lower bound of the character's estimated width, at least 1 pixel.
int	CharWidthUB [get, set]
	The upper bound of the character's estimated width, should be larger than or equal to the lower bound.
double	DotGap [get, set]
	The maximum distance between two nearby dots' centers. By default, it's 0 which means we'll decide it on the fly based on the distributions of all the distances.
BOOL	DotPrint [get, set]
	Whether to segment as the dot-print character. By default, it's no.
double	DotSize [get, set]
	The average area of individual dots, used to determine whether a foreground object is the character dot or some noisy fragment. By default, it's 0 which means we'll decide it on the fly based on the distributions of all foreground objects.
ILFontDatabase	FontDatabase [get, set]
	The associated font database.
int	FragmentThreshold [get, set]
	The maximum area of object that could be treated as fragment, then it will be merged to neighbors to form a valid character. By default, it's 0 which means we'll decide it on the fly based on the distributions of all foreground objects.
int	NoiseThreshold [get, set]
	The minimum area of valid character or fragment in pixel. By default, it's 0 which means we'll decide it on the fly based on the distributions of all foreground objects.
LPVPolarity	Polarity [get, set]
	The object polarity of the characters from the background.
LPVOCRSegmentFlags	SegmentFlags [get, set]
	The extra segmentation flags to control the line and character segmentation behavior.
wchar_t	UnknownPlaceholder [get, set]
	The place holder for unknown characters. By default, it's the question symbol '?'.
LPVOCVFlags	VerifyFlags [get, set]
	The verification flags to control which parts contribute to the verification score.

Detailed Description

This interface provide functionalities of character segmentation, recognition and verification.

The OCR tool is used for vision-based character segmentation, recognition and verification. It is supported to extract multiple lines of text by one scanning. It's also capable to separated ligatures in line and sticky lines, which is a common case in laser-printed labels. Recognition and verification is processed based on the specified font database and the trained character models inside it.

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

Example Code

C++

ILOCRPtr ocrTool = LOCR::Create();
ocrTool->FontDatabase = fontDb;
ocrTool->SetCharSize(100, 150, 20, 20); // 100 x 150 +- 20%
ocrTool->NoiseThreshold = 20;
ILOCRResultPtr ocrResult;
LPVErrorCode err = ocrTool->Detect(img, roi, &ocrResult);
if (err != LPVErrorCode::LPVNoError) return; // failed, or no result

C#

LOCR ocrTool = new LOCR();
ocrTool.FontDatabase = fontDb;
ocrTool.SetCharSize(100, 150, 20, 20); // 100 x 150 +- 20%
ocrTool.NoiseThreshold = 20;
ILOCRResultPtr ocrResult;
LPVErrorCode err = ocrTool.Detect(img, roi, out ocrResult);
if (err != LPVErrorCode.LPVNoError) return; // failed, or no result

COM

ILOCRPtr ocrTool = LOCR::Create();
// or, ILOCRPtr ocrTool(__uuidof(LOCR)), for absence of built-in Create functions (before 2.2.0)
ocrTool->FontDatabase = fontDb;
ocrTool->SetCharSize(100, 150, 20, 20); // 100 x 150 +- 20%
ocrTool->NoiseThreshold = 20;
ILOCRResultPtr ocrResult;
LPVErrorCode err = ocrTool->Detect(img, roi, &ocrResult);
if (err != LPVErrorCode::LPVNoError) return; // failed, or no result

Member Function Documentation

Detect()

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

Segment and recognize the lines and characters from the given image and region.

Parameters

[in]	img	The input image.
[in]	region	The input region. For the rotated rectangle region, the region's angle is used as the main direction of the text line. For the others, the main direction is the image's x axis.
[out]	results	Output the segmentation and recognition result.

Return values

error

Return error code if anything is wrong.

See also

Segment(), DetectAndVerify()

DetectAndVerify()

LPVErrorCode DetectAndVerify	(	ILImage *	img,
		ILRegion *	region,
		ILOCRResult **	results
	)

Segment, recognize and verify the lines and characters from the given image and region.

Parameters

[in]	img	The input image.
[in]	region	The input region. For the rotated rectangle region, the region's angle is used as the main direction of the text line. For the others, the main direction is the image's x axis.
[out]	results	Output the segmentation, recognition and verification result.

Return values

error

Return error code if anything is wrong.

See also

Segment(), Detect()

Segment()

LPVErrorCode Segment	(	ILImage *	img,
		ILRegion *	region,
		ILOCRResult **	results
	)

Segment the lines and characters from the given image and region.

Parameters

[in]	img	The input image.
[in]	region	The input region. For the rotated rectangle region, the region's angle is used as the main direction of the text line. For the others, the main direction is the image's x axis.
[out]	results	Output the segmentation result.

Return values

error

Return error code if anything is wrong.

See also

Detect(), DetectAndVerify()

SetCharSize()

void SetCharSize	(	int	w,
		int	h,
		double	wTolerance,
		double	hTolerance
	)

Set the character's estimated size range.

Parameters

[in]	w	The estimated width.
[in]	h	The estimated height.
[in]	wTolerance	The tolerance of width in percentage. It generates the width bounds as \( w /times (1 - wTol \%) ~ w /times (1 + wTol \%) \) .
[in]	hTolerance	The tolerance of height in percentage. It generates the height bounds as \( h /times (1 - hTol \%) ~ h /times (1 + hTol \%) \) .

SetCharSkew()

void SetCharSkew	(	double	skewLB,
		double	skewUB
	)

Set the skew angle range of the italic characters, in -60 ~ 60 degree.

Parameters

[in]	skewLB	The lower skew angle bound
[in]	skewUB	The upper skew angle bound