GSTFEATURE - Feature representation of a tensor field.

Contents

Description

Derive features like norm, orientation, coherence, inertia... from tensor representation.

Syntax

    f = GSTFEATURE(gx2, gy2, gxy);
    [f1, f2, ...] = GSTFEATURE(gx2, gy2, gxy, feature,
                   'Property', propertyvalue, ...);
    F = GSTFEATURE(gx2, gy2, gxy, feature, 'Property', propertyvalue, ...);

Inputs

gx2, gy2, gxy : components of a symmetric (structure) tensor T.

feature : optional string or cell of strings setting the feature to be estimated; it is one of the following:

or any cell combining those strings (order has no importance); default: feature='norm'.

Property [propertyname propertyvalues]

'eign' : optional string defining the method used for the tensor's norm approximation; it is either:

default: eign='l1'.

'ex', 'ey' : optional vectorial field provided by the vectors [ex,ey] used for computing the feature 'orvec'; default: ex=ey=[].

Output

f : matrices of features derived from the GST; in the case one output only is provided and several features are computed, the features are concatenated into the single output.

References

[Zenzo86] S. Di Zenzo: "A note on the gradient of a multi-image", CVGIP, 33:116-125, 1986. http://www.sciencedirect.com/science/article/pii/0734189X86902239

[Cuma91] A. Cumani: "Edge detection in multispectral images", CVGIP: Graphical Models and Image Processing, 53(1):40-51, 1991. http://www.sciencedirect.com/science/article/pii/104996529190018F

[VV95] L. van Vliet and P. Verbeek: "Estimators for orientation and anisotropy in digitized images", Proc. ASCI, pp. 442-450, 1995. http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.104.8552

[Kosch95] A. Koschan: "A comparative study on color edge detection", Proc. ACCV, pp. 574-578, 1995. http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.31.2648

[Cuma98] A. Cumani: "Efficient contour extraction in color images", Proc. ACCV, pp. 582-589, LNCS 1351, 1998. http://www.springerlink.com/content/c1750730487puqm0/

[Weick98] J. Weickert: "Anisotropic Diffusion in Image Processing", Teubner-Verlag, Stuttgart (Germany), 1998.

[Kothe03] U. Koethe: "Integrated edge and junction detection with the boundary tensor", Proc. IEEE ICCV, vol. 1, pp. 424?431, 2003. http://hci.iwr.uni-heidelberg.de/Staff/ukoethe/papers/polarfilters.pdf

[Scheun03] P. Scheunders: "A wavelet representation of multispectral images", Frontiers of Remote Sensing Information Processing, pp. 197?224. World Scientific, 2003. http://ebooks.worldscinet.com/ISBN/9789812796752/9789812796752_0009.html

[Tschum06] D. Tschumperle: "Fast anisotropic smoothing of multivalued images using curvature-preserving PDE?s", International Journal of Computer Vision, 68(1):65?82, 2006. http://www.springerlink.com/content/5217329131181uh4/

[WCFTA06] H. Wang, Y. Chen, T. Fang, J. Tyan and N. Ahuja: "Gradient adaptive image restoration and enhancement", Proc. ICIP, vol. 2, pp. 893-896, 2006. http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=4107174

See also

Ressembles: GSTSMOOTH, GRD2GST, GSTDECOMP. Requires: GSTFEATURE_BASE.

Function implementation

function varargout = gstfeature(gx2, gy2, gxy, varargin)

parsing parameters

error(nargchk(1, 18, nargin, 'struct'));
error(nargoutchk(1, 7, nargout, 'struct'));

if ~(isnumeric(gx2) && isnumeric(gy2) && isnumeric(gxy))
    error('gstfeature:inputerror','matrices are required in input');
end

p = createParser('GSTFEATURE');   % create an instance of the inputParser class.
p.addOptional('f','norm', @(x)ischar(x) || iscell(x));
% any(strcmpi(x,{'norm','frob','orient','ordir','reorient','inert','coher'});
p.addParamValue('eign',[],@(x)ischar(x) && ...
    any(strcmpi(x,{'abs','zen','l1','sap','sum','ndi','dif','koe'})));
p.addParamValue('ex',[],@(x)isnumeric(x));
p.addParamValue('ey',[],@(x)isnumeric(x));

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

internal variables and further testing

% check the compatibility of the input parameters
if any(size(gx2)~=size(gy2)) || any(size(gx2)~=size(gxy))
    error('gstfeature:inputerror','matrices must have same dimensions');
end

% % create the list of feature names
% if nb_dims(p.f)==2 % then already an array of feature names has been passed
%     lfeat = p.f;
%
% else   % let's create such an array from the concatenate string
%     lfeat = str2vsubstr(p.f, ...
%         strvcat('eigenorm','norm','frobenius','frob','orientation', ...
%         'orient','direction','ordir','vectorial','orvec', ...
%         'inertia','inert','coherence','coher'));                       %#ok
% end
if ischar(p.f),  p.f = {p.f};  end
nfeat = numel(p.f);

if nargout~=1 && nargout~=nfeat
    error('gstfeature:inputerror', ...
        ['the no. of output arguments must be 1 or equal to the no. ' ...
        'of computed features must be equal']);
end

if ~any(strncmp('norm',p.f,4)) && ~any(strncmp('eigenorm',p.f,4)) && ...
        ~isempty(p.eign)
        warning('gstfeature:incompatibleparameter',...
            'parameter eign ignored for features other than norm');
else
    p.eign = 'zen'; % default value, may be unused
end

test = ~any(strncmp('orvec',p.f,4)) && ~any(strncmp('vectorial',p.f,4));
if test && (~isempty(p.ex) || ~isempty(p.ey))
    warning('gstfeature_base:incompatibleparameter',...
        'parameters ex,ey ignored for features other than orvec');
elseif ~test && isempty(p.ex) && isempty(p.ey)
    error('gstfeature_base:incompatibleparameter',...
        'parameters ex,ey need to be passed with feature orvec');
end

main computation

varargout{1} = [];

for i=1:nfeat
    F = gstfeature_base(gx2, gy2, gxy, p.f{i}, p.eign, p.ex, p.ey);
    if nargout==1,  varargout{1} = cat(3, varargout{1}, F);
    else            varargout{i} = F;                        end
end

display

if p.disp
    figure,
    ncols = min(3,nfeat); nrows = ceil(nfeat/ncols);
    for i=1:nfeat
        subplot(nrows,ncols,i), imagesc(varargout{i}), axis image off;
        title(p.f(i,:)), colormap gray;
    end
end

end % end of gstfeature