Randomized Hough transform (RHT) and Genetic Algorithm (GA)

Hi everybody,

I need to detect the edges in the following image, I used RHT (Randomized Hough transform) for detecting curves.

Here is the input (original) image:

Here is my code (RHT):

clear all
close all
clc

i = imread ('d (457).bmp');
ig = double(rgb2gray(i));

figure
imshow(ig, [min(min(ig)) max(max(ig))]);

h1 = fspecial('sobel');
h2 = h1';
igh = imfilter(ig,h1);
igv = imfilter(ig,h2);
igs = abs(igh)+abs(igv);

figure;
imshow(igs, [min(min(igs)) max(max(igs))]);
igsT = igs>110;
figure;
imshow(igsT);


acc = ([size(ig) 21]);
[r,c]=find(igsT);

iter = 0;
while (iter<200000)
    iter = iter+1;
    N = length(r);
    ind = floor(N*rand(1,3))+ 1;
    while(length(unique(ind))<3)
        ind = floor(N*rand(1,3))+1;
    end
%     [x0,y0,R] = threepoints([r(ind(1)) c(ind(1))], [r(ind(2)) c(ind(1))], [r(ind(3)) c(ind(3))]);
    x0 = [r(ind(1)) c(ind(1))];
    y0 = [r(ind(2)) c(ind(1))];
    R = [r(ind(3)) c(ind(3))];
    x0 = ceil(x0);
    y0 = ceil(y0);
    R = ceil(R);
    if (ismember(x0, [1 size(ig,1)]))
        if (ismember(y0, [1 size(ig,2)]))
            if(ismember(R, [15 25]))
                acc(x0, y0, R-14) = acc(x0,y0,R-14)+1;
                if(acc(x0,y0,R-14)>4)
                    s = fprintf('Found circle with [x0, y0, R] = [%d %d %d], press any key to continue\n',x0, y0,R);
                    disp(s)
                    hold on
                    cc = circle([y0, x0], R, 20,'-');
                    pause
                   
                end
            end
        end
    end
end

RGB_Image = uint8( igsT(:,:,[1 1 1]) * 255 );
figure;
imshow(RGB_Image);

After running this code, I got the following result for RHT (for input original bmp image):

Whether this result is satisfactory or I need to increase the value of the threshold (currently 110)?

Also, i would like to ask if anyone has idea how RHT could be combined with a genetic algorithm (GA) in MATLAB, in order to get more efficient results compared to using only (RHT) or only GA (based on edge linking for edge detection)?