About fourier descriptors

Hi , is the following code right for getting fourier descriptors for a shape ?

[x3,y3] = find(im3,1);
[x4,y4] = find(im4,1);
b3 = bwtraceboundary(im3,[x3 , y3],'N',8,nnz(im3),'clockwise');
b4 = bwtraceboundary(im4,[x4 , y4],'N',8,nnz(im4),'clockwise');
f3 = frdescp(b3);
f4 = frdescp(b4);
f3 = abs(f3);
f4 = abs(f4);
f3 = f3(2:10);
f4 = f4(2:10);
f3 = f3/f3(1);
f4 = f4/f4(1);

is that true ? providing that when i plotted the two boundaries i got those two figures

http://i46.tinypic.com/x5nrk7.png

http://i50.tinypic.com/oi9h8k.png

Note that f3 has the following values

    1.0000
    0.6550
    0.4776
    0.6454
    0.1478
    0.8498
    0.4893
    0.9819
    0.7874

and f4

    1.0000
    2.0077
    0.4963
    0.3795
    0.5872
    1.0960
    0.8823
    0.8092
    0.8482

Both pics have the same size as well

So what is wrong ?

[Walter Roberson]

What difference do you see between what you get and what you expect?

Answer 1

[Hayan Tariq]

I am also trying to use fourier descriptor. why do you think there is something wrong. does the descriptor retrieve back almost the original shape? if yes then it is ok.

I am trying to use fourier descriptors as a feature for a classifier like neural net. but i don't know how to use it. can I use like the first 10 descriptor like how you extract the first 10, just use them as an input or there is another way?

thanks