iir_coeff.h File Reference

#include <cmath>
#include <complex>
#include <iostream>
#include <Halide.h>

Go to the source code of this file.

Functions
std::vector< float >	overlap_feedback_coeff (std::vector< float > a, std::vector< float > b)
	Compute the coefficients of a higher order filter that is equivalent to two cascaded lower order filters of given coefficients. More...
std::vector< float >	integral_image_coeff (int iterations)
	Feed forward and feedback coefficients for computing n integrals of image, single intergal is summed area table and multiple integrals are multiple applications of summed area table. More...
std::vector< float >	gaussian_weights (float sigma, int order)
	Wrapper to compute third order recursive filter weights for Gaussian blur. More...
int	gaussian_box_filter (int iterations, float sigma)
	Compute the size of a box filter that approximates a Gaussian. More...
template<typename T >
Halide::Image< T >	reference_gaussian (Halide::Image< T > in, T sigma)
	Apply Gaussian filter on an input image. More...
Compute the Gaussian, derivative of Gaussian and integral of Gaussian
Parameters [in] x input (float or Halide::Expr) [in] mu mean of the Gaussian function [in] sigma sigma support of the true Gaussian filter
float	gaussian (float x, float mu, float sigma)
float	gaussDerivative (float x, float mu, float sigma)
float	gaussIntegral (float x, float mu, float sigma)
Halide::Expr	gaussian (Halide::Expr x, float mu, float sigma)
Halide::Expr	gaussDerivative (Halide::Expr x, float mu, float sigma)
Halide::Expr	gaussIntegral (Halide::Expr x, float mu, float sigma)

Function Documentation

std::vector<float> overlap_feedback_coeff	(	std::vector< float >	a,
		std::vector< float >	b
	)

Compute the coefficients of a higher order filter that is equivalent to two cascaded lower order filters of given coefficients.

Todo:

This function is only partially tested and may be unstable

Parameters

[in]	a	coefficients of first lower order filter
[in]	b	coefficients of second lower order filter

Returns

coefficients of higher order filter

std::vector<float> integral_image_coeff

(

int

iterations

)

Feed forward and feedback coefficients for computing n integrals of image, single intergal is summed area table and multiple integrals are multiple applications of summed area table.

Parameters

[in]

iterations

number of integrals

Returns

coeff of IIR filter that computes multiple integrals

float gaussian	(	float	x,
		float	mu,
		float	sigma
	)

Referenced by reference_gaussian().

float gaussDerivative	(	float	x,
		float	mu,
		float	sigma
	)

float gaussIntegral	(	float	x,
		float	mu,
		float	sigma
	)

Halide::Expr gaussian	(	Halide::Expr	x,
		float	mu,
		float	sigma
	)

Halide::Expr gaussDerivative	(	Halide::Expr	x,
		float	mu,
		float	sigma
	)

Halide::Expr gaussIntegral	(	Halide::Expr	x,
		float	mu,
		float	sigma
	)

std::vector<float> gaussian_weights	(	float	sigma,
		int	order
	)

Wrapper to compute third order recursive filter weights for Gaussian blur.

Third order filter can be approximated by cascaded first and second order filters

Parameters

[in]	order	order of recursive filter for approximation
[in]	sigma	sigma of Gaussian filter

Returns

vector with feedforward coeff as first element and rest feedback coeff

int gaussian_box_filter	(	int	iterations,
		float	sigma
	)

Compute the size of a box filter that approximates a Gaussian.

Source: "Efficient Approximation of Gaussian Filters" Rau and McClellan, IEEE Trans. on Signal Processing 1997 http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=554310

Parameters

[in]	iterations	number of repeated applications of box filter
[in]	sigma	sigma support of the true Gaussian filter

Returns

box filter width

template<typename T >

Halide::Image<T> reference_gaussian	(	Halide::Image< T >	in,
		T	sigma
	)

Apply Gaussian filter on an input image.

Parameters

[in]	in	input single channel image
[in]	sigma	sigma support of the true Gaussian filter

Returns

filtered image

Definition at line 80 of file iir_coeff.h.

References gaussian().