CONVOLUTION - Convolution based filtering.

Contents

Description

Compute convolution with centered filter.

Syntax

   y = CONVOLUTION(x, h);
   y = CONVOLUTION(x, h, bound);

Inputs

x : input signal; either 1D or 2D.

h : centered filtered used to convolve x; it is centered at 0 for odd length of the filter, and at 1/2 otherwise.

bound : optional string parameter; it is either 'per' for periodic extension of the signal or 'sym' for symmetric extension.

Output

y : convolved signal.

See also

Ressembles: DIRGAUSSKERNEL, GAUSSKERNEL, HOURGLASSKERNEL, CONV, CONV2, IFFT, FFT, FFT2. Requires: CONVOLUTION_BASE.

Function implementation

function y = convolution(x, h, varargin)

parsing parameters

error(nargchk(1, 11, nargin, 'struct'));
error(nargoutchk(1, 1, nargout, 'struct'));

if ~isnumeric(x)
    error('convolution:inputerror','numeric signal required in input');
end

p = createParser('CONVOLUTION');   % create an instance of the inputParser class.
p.addOptional('bound','sym', @(x)ischar(x) && any(strcmpi(x,{'sym','per'})));

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

main computation

[y,nd] = convolution_base(x, h, p.bound);

display

if p.disp
    figure,
    if nd==1,    plot(y,'r-');
    else
        imagesc(rescale(y,0,1)), axis image off;
        if size(y,3)==1,  colormap gray;  end
    end
    title('convolved signal');
end

end % end of convolution