You are here: Start » Function Reference » Image » Image Enhancement » NormalizeImage

NormalizeImage

Header:	AVL.h
Namespace:	avl
Module:	FoundationLite

Rescales an image linearly, so that its minimum becomes inNewMinimum and the maximum of the remaining pixels becomes inNewMaximum.

Applications: Aims at better using the image's dynamic range to represent an interesting subset of pixel values.

Syntax

void avl::NormalizeImage
(
	const avl::Image& inImage,
	atl::Optional<const avl::Region&> inRoi,
	float inNewMinimum,
	float inNewMaximum,
	float inSaturateBrightestFraction,
	float inSaturateDarkestFraction,
	atl::Optional<float> inMinValue,
	atl::Optional<float> inMaxValue,
	avl::Image& outImage,
	float& outA,
	float& outB,
	avl::Region& diagLinearNormalizedRegion
)

Parameters

Name	Type	Range	Default	Description
inImage	const Image&			Input image
inRoi	Optional<const Region&>		NIL	Range of pixels to be processed
inNewMinimum	float		0.0f	Desired minimum value of the resulting image
inNewMaximum	float		255.0f	Desired maximum value of the resulting image
inSaturateBrightestFraction	float	0.0 - 1.0	0.0f	Fraction of the brightest pixels skipped during normalization
inSaturateDarkestFraction	float	0.0 - 1.0	0.0f	Fraction of the darkest pixels skipped during normalization
inMinValue	Optional<float>		NIL	Pixels darker than that value will be skipped during normalization
inMaxValue	Optional<float>		NIL	Pixels brighter than that value will be skipped during normalization
outImage	Image&			Rescaled image
outA	float&			Multiplicative parameter of the applied linear transformation of pixel values
outB	float&			Additive parameter of the applied linear transformation of pixel values
diagLinearNormalizedRegion	Region&			Region of image that has been linearly normalized

Description

This filter linearly scales the pixel values of an image in order to make the histogram span the desired range of values.

The operation computes the parameters \(A\), \(B\) of the linear transform that scales the image values to the desired range and applies the transform computing the results as follows:

\[ \begin{aligned} outImage[i,j] &= inImage[i,j] \times A + B \\ outA &= A \\ outB &= B \end{aligned} \]

The inSaturateBrightestFraction and inSaturateDarkestFraction parameters can be used to make image normalization independent from salt and pepper noise. The normalization skips a chosen fraction of the brightest and the darkest pixels during counting \(A\) and \(B\). The brightest and darkest pixels are set to inNewMaximum and inNewMinimum respectively, while the rest of the image is linearly scaled. A region of the linear scaling is available on diagLinearNormalizedRegion diagnostic output. For example, setting inSaturateBrightestFraction to 0.01 causes skipping 1 percent of the brightest pixels during counting \(A\) and \(B\). Please note that the inSaturateBrightestFraction and inSaturateDarkestFraction parameters can be non-zero only for images with UInt8, Int8, UInt16 or Int16 pixel type.

Examples

NormalizeImage run on example image.

Errors

List of possible exceptions: