Prev: To draw the major and minor axis of an ellipse
Next: matrix inverse using qr decomposition
From: Adam Bartoszewski on 3 Jan 2010 13:34
i think u r wrong at the beginning
the centroid u r calculating should be calcualted from only the shape not the whole image.
But anyway if u have fixed the problem i d apriciate if u could share the final code, because now im starting to code the GFD and also have some dificulties
thanks in advance

"shantini jey" <brutal_setan(a)hotmail.com> wrote in message <gp4jmi$ai2$1(a)fred.mathworks.com>...
> hi, im doin shape based image retrieval using generic fourier descriptor..
> the methods to generating GFD are:
> 1. Input shape image data f(x, y);
> 2. Get centroid of the shape (xc, yc);
> 3. Set the centroid as the origion; /* translation normalization */
> 4. Get the maximum radius of the shape image (maxRad);
> 5. Polar Fourier transform
> 6. Calculate FD
> 7. Output feature vector FD.
>
> i've done some work on it.. but somehow the FD part is not correct..
> pls some1 help me... this is my coding...
>
> if nargin < 1
> error('Please load an image');
> end
>
> img = double(img);
> [rows,cols] = size(img);
> cy = round(rows/2);
> cx = round(cols/2);
>
> if exist('radius','var') == 0
> rmax= min(round(rows/2),round(cols/2))-1;
> end
> radius = sqrt((cx-(rmax^2)) + (cy-(rmax^2)));
>
> if exist('angle','var') == 0
> angle = 360;
> end
>
> pcimg = [];
> i = 1;
>
> for r=0:rmax
> j = 1;
>
> for a=0:2*pi/angle:2*pi-2*pi/angle
> pcimg(i,j) = img(cy+round(r*sin(a)),cx+round(r*cos(a)));
> j = j + 1;
> end
> i = i + 1;
> end
>
>
> %------------------------------------------------------
>
> %FD
>
> function FD = frdescp(pcimg)
>
> FR= pcimg(i,j)*cos(2*pi*r*(radius/rmax) + angle*a); %real part of spectra */
> FI= pcimg(i,j)*sin(2*pi*r*(radius/rmax) + angle*a); %imaginary part of spectra */
>
>
>
> if (angle==0 && radius==0)
>
> FD(0) = sqrt((FR^2+ FR^2)/(pi*maxRad^2));
> Else
> FD(r+ang) = sqrt((FR^2*(r*angle) + FI^2*(radius*angle))/ FD(0));
> end
> n can some1 pls explain wat is output feature vector FD..
> thankz a million in advance
 | 
Pages: 1
Prev: To draw the major and minor axis of an ellipse
Next: matrix inverse using qr decomposition