Fitting spline to outline

I have been trying to write a code (shown below) that will identify the outline of an object and then fit a spline to it.

I can identify the edge (using a ball for example below, image 1), but it picks up multiple points around the edge for each x and y value (image 2). So, I've tried to clean it up be removing some of the points.

However, the final spline doubles back on itself (image 4). How do I stop that from happening and just plot a smooth line connecting all the points?

% Get outline
BW = edge(Image, 'Canny',  0.6);
B = bwboundaries(BW);
J = num2cell(cell2mat(B),1);
F(:,1) = J{1,1};
F(:,2) = J{1,2};

%Show image and outline points
figure
imshow(I)
hold on
scatter(F(:,1),F(:,2), 'r0')

%Reduce number of points
for q = 20:20:(numel(F)/2)
    F1((q/20),:) = F(q,:);
end

% Fit spline
x = F1(:,1);
y = F1(:,2);
pointspline = [x'; y'];
values = cscvn(pointspline);
sp = fn2fm(values,'B-')

%Show reduced points and spline
figure
scatter(F1(:,1),F1(:,2), 'r')
hold on
fnplt(sp)

1.

2.

3.

4.

[darova]

If the curve double back can you just use a half of data?

%Reduce number of points
% for q = 20:20:(numel(F)/2)
%     F1((q/20),:) = F(q,:);
% end

    % can be written as:
F1 = F(1:20:end/2,:);

[Mehri Mehrnia]

something is wrong with this code, I used it for a boundary of image. it shifts it 90 degrees.

Answer 1

[Pujitha Narra]

Hi John,

There is a similar example here:

https://in.mathworks.com/help/images/boundary-tracing-in-images.html

Hope this helps.