How do I develop a pseudo Zernike Moments proposed by Al-Rawi, 2010?

1 visualización (últimos 30 días)
Aj_ti el 12 de Jun. de 2016
Abierta de nuevo: Aj_ti el 13 de Jun. de 2016
Based on the file shared here : fast computation of PZM by Sadeq al-Rawi,how do I continue the work to compute the moments based on his proposed method (mentioned in the paper 'Fast computation of pseudo Zernike moments')? I'd tried to build full code but it give me poor performance (in term of accuracy) when I implement it into face recognition system. EDIT: MY CODE IS AS FOLLOW:
img1 = imread('exp.bmp');
N = 4; %order
img =double(img1);
S = size(img, 1);
xstep = 2/(S-1);
[X ,Y] = meshgrid(0:xstep:1,0:xstep:1);
Rd = sqrt((2.*X-S+1).^2+(S-1-2.*Y).^2)/S;
theta = atan((S-1-2.*Y)/ (2.*X-S+1));
Rd = (Rd<=1).*Rd;
Rd =[Rd(:)];
theta =[theta(:)];
Rad = pseudo_zernike_radial_polynomials(N, Rd);
op =[];
vectr = [];
prodc = [];
PZm = [];
%algorithm 2 as in al-rawi, 2010
psi = [];
arc = atan((S/2-10)/(S-20));
a = S-20;
b = S/2-10;
na = 20;
nb = S/2+10;
ay = 20;
nay = S-20;
bx = S/2+10;
nbx = S/2-10;
for m=0:N
if mod(m,2)==0
x1 = img(a,b)+img(na, nb);
x2 = img(a, nb) +img(na,b);
y1 = ((-1)^m/2)*(img(nbx, ay)+img(bx, nay));
y2 = ((-1)^m/2)*(img(bx, ay)+img(nbx, nay));
psii = cos (m*arc)*(x1+y1+x2+y2)-1j*sin(m*arc)*(x1+y1-x2-y2);
x1 = img(a,b)-img(na, nb);
x2 = img(a, nb) -img(na,b);
y1 = ((-1)^m/2)*(img(nbx, ay)-img(bx, nay));
y2 = ((-1)^m/2)*(img(bx, ay)-img(nbx, nay));
psii = cos(m*arc)*(x1+x2)+sin(m*arc)*(y1-y2)+1j*(cos(m*arc)*(y1+y2)-sin(m*arc)*(x1-x2));
psi =[psi,psii];
for ord = 0:N
for lpsi =1:length(psi)
prod = psi(lpsi).*Rad(ord+1,:)';
prodc = [prodc, prod];
PZmn = ((N+1)/pi)*sum(sum(prodc(:, ord+1)));
PZm = [PZm; PZmn];

Respuestas (0)


Más información sobre Zernike Polynomials en Help Center y File Exchange.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by