ILFontDatabase Interface Reference

This interface represents a font database for character recognition. More...

Inheritance diagram for ILFontDatabase:

Public Member Functions
void	AddOCRChar (ILOCRChar *charData, LString charString)
void	AddOCRLine (ILOCRLine *lineData, LString lineString)
void	AddOCRResult (ILOCRResult *resultData, LString resultString)
int	CountChar ()
ILSampleDatabase *	CreateChar (LString charString)
ILSampleDatabase *	GetCharDatabase (LString charString)
void	Init (LString fontName, LString workingDirectory)
BOOL	IsInMemory ()
	Whether this is an in-memory database, whose sample files and all informations are never saved.
BOOL	IsTrained ()
LString	ItemChar (int index)
void	RemoveChar (LString charString)
LPVErrorCode	Train ()
Public Member Functions inherited from ILObject
ILObject *	Copy ()
LPVErrorCode	Load (LString filename)
void	Reset ()
LPVErrorCode	Save (LString filename)
BOOL	Valid ()

Properties
BOOL	AutoFeature [get, set]
	Whether to enable automatic feature selection and tuning. By default, it's off.
BOOL	AutoModel [get, set]
	Whether to enable automatic model selection and tuning. By default, it's off.
LPVOCRFeatures	Features [get, set]
	The features used for recognition. By default, it's LPVOCRFeatureDefault. More...
LPVModelType	ModelType [get, set]
	The machine learning model type used for recognition.
LString	Name [get]
	The name of the sample database, which is usually unique and as an identification for the database globally.
LSize	PixelGrid [get, set]
	The grid size for LPVOCRFeaturePixel feature. By default, it's a \( 6 \times 8 \) grid.
LPVOCRPixelProjection	PixelProjection [get, set]
	The grid projection used for LPVOCRFeaturePixel feature. By default, it's LPVOCRPixelProjNone indicating no projection.
LPVOCRPixelValue	PixelValue [get, set]
	The pixel value's calculation method used for LPVOCRFeaturePixel feature. By default, it's LPVOCRPixelGrayscaleNormalized indicating normalized grayscale value.
LString	WorkingDir [get]
	The working directory of the font database. For the persisted database, it is used to store the sample files and management information file. For the in-memory database, it is the fix "in-memory" text.

Detailed Description

This interface represents a font database for character recognition.

The font database is used to manage the character samples, and train the machine-learning based classification model for recognition. It provides various image-based or geometric features designed for OCR task. It also support advanced automatic feature selection and model tuning, if there are sufficient character samples. It could be serialized to disk and shared by multiple OCR tools.

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

Example Code

C++

ILFontDatabasePtr fontDb = LFontDatabase::Create();
fontDb->Init(L"ABC Font", L"D:/MyFontDB/");
// add samples
fontDb->AddOCRResult(segResult, L"ABC\nDEF\n123");
fontDb->AutoFeature = true; // auto-selected
LPVErrorCode err = fontDb->Train();
if (err != LPVErrorCode::LPVNoError) return;
// set the font database to OCR tool
ocrTool->FontDatabase = fontDb;

C#

LFontDatabase fontDb = new LFontDatabase();
fontDb.Init("ABC Font", "D:/MyFontDB/");
// add samples
fontDb.AddOCRResult(segResult, "ABC\nDEF\n123");
fontDb.AutoFeature = true; // auto-selected
LPVErrorCode err = fontDb.Train();
if (err != LPVErrorCode.LPVNoError) return;
// set the font database to OCR tool
ocrTool.FontDatabase = fontDb;

COM

ILFontDatabasePtr fontDb = LFontDatabase::Create();
// or, ILFontDatabasePtr fontDb(__uuidof(LFontDatabase)), for absence of built-in Create functions (before 2.2.0)
fontDb->Init(L"ABC Font", L"D:/MyFontDB/");
// add samples
fontDb->AddOCRResult(segResult, L"ABC\nDEF\n123");
fontDb->AutoFeature = VARIANT_TRUE; // auto-selected
LPVErrorCode err = fontDb->Train();
if (err != LPVErrorCode::LPVNoError) return;
// set the font database to OCR tool
ocrTool->FontDatabase = fontDb;

Member Function Documentation

AddOCRChar()

void AddOCRChar	(	ILOCRChar *	charData,
		LString	charString
	)

Add the single character object to the font database as training sample.

Parameters

[in]	charData	The single character object.
[in]	charString	The recognition string. Pass in empty to use the recognition result inside charString as training characters.

See also

AddOCRResult(), AddOCRLine()

AddOCRLine()

void AddOCRLine	(	ILOCRLine *	lineData,
		LString	lineString
	)

Add all the characters in the line of text object to the font database as training samples.

Parameters

[in]	lineData	The line of text object. It should contain at least one character.
[in]	lineString	The recognition string. It should match the count of the characters in lineData. Pass in empty to use the recognition result inside lineData as training characters.

See also

AddOCRResult(), AddOCRChar()

AddOCRResult()

void AddOCRResult	(	ILOCRResult *	resultData,
		LString	resultString
	)

Add all the characters in the OCR result object to the font database as training samples.

Parameters

[in]	resultData	The OCR result object. It should contain at least one line or one character.
[in]	resultString	The recognition string. It should match the count of the characters in resultData. Pass in empty to use the recognition result inside resultData as training characters.

See also

AddOCRLine(), AddOCRChar()

CountChar()

int CountChar

(

)

Get the total count of all the characters.

Return values

val	The total count.

See also

CreateChar(), RemoveChar(), ItemChar(), GetCharDatabase()

CreateChar()

ILSampleDatabase* CreateChar

(

LString

charString

)

Create a new character in the font database. It will also create a sample database under the font database for the new character at the same time.

Parameters

[in]

charString

The character code.

Return values

charDB

Return the sample database for the new character

See also

RemoveChar(), CountChar(), ItemChar(), GetCharDatabase()

GetCharDatabase()

ILSampleDatabase* GetCharDatabase

(

LString

charString

)

Get the corresponding database by the given character code.

Parameters

[in]

charString

The character code

Return values

charDB

Return the sample database for the character

See also

CreateChar()

Init()

void Init	(	LString	fontName,
		LString	workingDirectory
	)

Create and initialized the font database.

Parameters

[in]	fontName	The name of the font database.
[in]	workingDirectory	The working directory. To create an in-memory database, pass in an empty path. For other cases, we'll create a persisted database and initialize the working directory with an empty database. So you should provide a readable and writable path. If there's an existing database in the given directory, we'll try to load it in the first place, and the label name may be refreshed with the loaded one.

IsTrained()

BOOL IsTrained

(

)

Check whether the font database is well-trained

Return values

val	Return True if it's trained, otherwise, return False.

ItemChar()

LString ItemChar

(

int

index

)

Fetch one character with given index.

Parameters

[in]

index

The input index, it should be a 0-based number smaller than total count.

Return values

result

Return the character code of given index.

See also

CreateChar(), RemoveChar(), CountChar(), GetCharDatabase()

RemoveChar()

void RemoveChar

(

LString

charString

)

Remove the character from the font database. Please noted the corresponding database is not permanently removed, you may delete it manually.

Parameters

[in]

charString

The character code to remove.

See also

CreateChar(), CountChar(), ItemChar(), GetCharDatabase()

Train()

LPVErrorCode Train

(

)

Train the font database. If automatic feature or model selection is turned on, it's performed during the training.

Return values

error

Return error code if anything is wrong.

Property Documentation

Features

LPVOCRFeatures Features

getset

The features used for recognition. By default, it's LPVOCRFeatureDefault.

/*!