This class implements functions to determine the frame variance. More...

Static Public Member Functions
template<typename T , typename TIntegral >
static bool	deviation1Channel8Bit (const T frame, const TIntegral integral, uint8_t *deviation, const unsigned int width, const unsigned int height, const unsigned int framePaddingElements, const unsigned int deviationPaddingElements, const unsigned int window)
	This functions determines the frame deviation of a 1 channel 8 bit frame. More...

template<typename T >
static bool	deviation1Channel8Bit (const T frame, uint8_t deviation, const unsigned int width, const unsigned int height, const unsigned int framePaddingElements, const unsigned int deviationPaddingElements, const unsigned int window)
	This functions determines the deviation within a 1-channel 8 bit frame. More...

template<typename TElementType , typename TSummationType , typename TMultiplicationType , unsigned int tChannels>
static void	imageStatistics (const TElementType frame, const unsigned int width, const unsigned int height, const unsigned int framePaddingElements, double mean=nullptr, double variance=nullptr, double standardDeviation=nullptr)
	Compute the per-channel mean, variance, and standard deviation over an image. More...

Detailed Description

This class implements functions to determine the frame variance.

Member Function Documentation

◆ deviation1Channel8Bit() [1/2]

template<typename T , typename TIntegral >

static bool Ocean::CV::FrameVariance::deviation1Channel8Bit	(	const T *	frame,
		const TIntegral *	integral,
		uint8_t *	deviation,
		const unsigned int	width,
		const unsigned int	height,
		const unsigned int	framePaddingElements,
		const unsigned int	deviationPaddingElements,
		const unsigned int	window
	)

static

This functions determines the frame deviation of a 1 channel 8 bit frame.

Parameters

frame	Original frame to determine the variance for
integral	Bordered integral frame of the original frame with border size equal have of the specified window size, the border must be mirrored
deviation	Resulting deviation frame
width	The width of the original frame
height	The height of the original frame
framePaddingElements	The number of padding elements at the end of each frame row, in elements, with range [0, infinity)
deviationPaddingElements	The number of padding elements at the end of each deviation row, in elements, with range [0, infinity)
window	Size of the window must be odd, with range [1, infinity)

Returns: True, if succeeded

◆ deviation1Channel8Bit() [2/2]

template<typename T >

static bool Ocean::CV::FrameVariance::deviation1Channel8Bit	(	const T *	frame,
		uint8_t *	deviation,
		const unsigned int	width,
		const unsigned int	height,
		const unsigned int	framePaddingElements,
		const unsigned int	deviationPaddingElements,
		const unsigned int	window
	)

static

This functions determines the deviation within a 1-channel 8 bit frame.

The function uses an integral image for the calculation of the deviation.

Parameters

frame	The frame for which the deviation will be determined, must be valid
deviation	Resulting deviation frame
width	The width of the given frame, in pixel, with range [1, infinity)
height	The height of the given frame, in pixel, with range [1, infinity)
framePaddingElements	The number of padding elements at the end of each frame row, in elements, with range [0, infinity)
deviationPaddingElements	The number of padding elements at the end of each deviation row, in elements, with range [0, infinity)
window	Size of the window must be odd, with range [1, min(width, height) * 2]

Returns: True, if succeeded

Template Parameters

T	The data type of the frame elements, either 'int8_t' or 'uint8_t'

◆ imageStatistics()

template<typename TElementType , typename TSummationType , typename TMultiplicationType , unsigned int tChannels>

void Ocean::CV::FrameVariance::imageStatistics	(	const TElementType *	frame,
		const unsigned int	width,
		const unsigned int	height,
		const unsigned int	framePaddingElements,
		double *	mean = `nullptr`,
		double *	variance = `nullptr`,
		double *	standardDeviation = `nullptr`
	)

static

Compute the per-channel mean, variance, and standard deviation over an image.

The basic outline of the computation is this (this example assumes no multi-channel and no stride but the actual implementation supports both, of course):

T0 imagePixels[pixelCount] = { ... };
T1 sum = 0;
T1 squareSum = 0;
for (i = 0; i < pixelCount; ++i)
{
    sum[i] += T1(imagePixels[i]);
    squareSum[i] += T1(T2(imagePixel[i]) * T2(imagePixels[i]))
}
double mean = (double)sum / (double)pixelCount
double variance = ((double)squareSum / pixelCount) - (mean * mean);
double stddev = std::sqrt(variance)
 
// Type `T0` is the type of the pixel elements, `T1` is used for the summations and type `T2` is used for multiplications.
// This allows for a optimal control over how to compute and with what precision.
// Example configurations are:
// - `T0`=float,         `T1`=float,              `T2`=double
// - `T0`=unsigned char, `T1`=unsigned long long, `T2`=unsigned short

Note: The most related OpenCV function is cv::meanStdDev()

Parameters

frame	The input frame; mustn't be empty and must be a 1-channel image
width	The width of the input image, range: [1, infinity)
height	The height of the input image, range: [1, infinity)
framePaddingElements	Optional number of padding elements at the end of each source row, in elements, with range [0, infinity)
mean	Array for the resulting per-channel mean values, will be ignored if set to `nullptr` otherwise it must have `channels` elements
variance	Array for the resulting per-channel variance, will be ignored if set to `nullptr` otherwise it must have `channels` elements
standardDeviation	Array for the resulting per-channel standard deviations, will be ignored if set to `nullptr` otherwise it must have `channels` elements

Template Parameters

TElementType	Type of the elements of the pixels of the input image
TSummationType	Type used for the internal computation of the pixel sums (should be at least as large as `TElementType`), cf. function description for details
TMultiplicationType	Type used to compute the square values of pixels (allows control over integer vs. floating point multiplication), cf. function description for details
tChannels	Number of channels of the input image, range: [1, infinity)

The documentation for this class was generated from the following file:

FrameVariance.h

Static Public Member Functions

Detailed Description

Member Function Documentation

◆ deviation1Channel8Bit() [1/2]

◆ deviation1Channel8Bit() [2/2]

◆ imageStatistics()