FUZZIMPULSE - Fuzzy impulse noise reduction.

Contents

Description

Perform the fuzzy impulse noise reduction in images following the grayscale approach of [SWNWK07].

Syntax

   F = FUZZIMPULSE(I);
   [F, N] = FUZZIMPULSE(I, K, L);

Inputs

I : input image, possibly multispectral (then, channels are processed separately).

K : half-size of the analyzing window used for impulse noise detection (see Section 2.1 of [SWNWK07], Eq.(2)); default: K=1.

L : half-size of the analyzing window used for input image filtering (see Section 3 of [SWNWK07], Eq.(9)); default: L=1.

Outputs

F : output filtered image (impulse noise reduced).

N : (optional) boolean image set to true for pixels where impulse noise has been detected.

Remark

In the case of multispectral inputs, this function applies uncorrelated filtering (channel by channel), thus the filtered output may present artifact values.

References

[SWNWK06] S. Schulte, V. de Witte, M. Nachtegael, D. van der Weken and E. Kerre: "Fuzzy two-step filter for impulse noise reduction from color images", IEEE. Trans. on Image Processing, 15(11):3568-3579, 2006. http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1709999&tag=1

[SWNWK07] S. Schulte, V. de Witte, M. Nachtegael, D. van der Weken and E. Kerre: "Fuzzy random impulse noise reduction method", Fuzzy Sets and Systems, 158(3):270-283, 2007. http://www.sciencedirect.com/science/article/pii/S0165011406004179

See also

Ressembles: RAMPSHARP. Requires: FUZZIMPULSE_BASE.

Function implementation

function [F,varargout] = fuzzimpulse(I, varargin)

error(nargchk(1, 13, nargin, 'struct'));
error(nargoutchk(1, 2, nargout, 'struct'));

if ~isnumeric(I)
    error('fuzzimpulse:inputparameter','matrix required in input');
end

parsing parameters

p = createParser('FUZZIMPULSE');
p.addOptional('K',1, @(x)isscalar(x) && x>=1 && x==round(x));
p.addOptional('L',1, @(x)isscalar(x) && x>=1 && x==round(x));

% parse and validate all input arguments
p.parse(varargin{:});
p = getvarParser(p);

main computation

[F,N] = fuzzimpulse_base(I, p.K, p.L);

if nargout==2,  varargout{1} = N;  end; % N computed at no cost

display

if p.disp
    figure,
    subplot(1,2,1), imagesc(N), axis image off, title('detected impulse noise'),
    subplot(1,2,2), imagesc(rescale(F)), axis image off, title('denoised output');
    if size(F,3)==1,  colormap gray;  end
end

end % end of fuzzimpulse