Connecting points on opposite end of mask

Given an mask image i have certain points as seen in the image, I want to connect the point to the point at the opposite side of it given the list of points/array of point.Well the output would look something like.

[Matt J]

Please provide a .mat file with the clean input image and the points.

[Catalytic]

Are the points given to us sorted in clockwise or counter-clockwise order?

[kunal]

there are some point clustered at some places, but i am working on it to get any one relevant point from them

[Matt J]

there are some point clustered at some places, but i am working on it to get any one relevant point from them

How about you do that first, and then get back to us. It will be much easier at that point.

[kunal]

@Catalytic They originate from a single point, traverse the mask in a convolution-like path, and loop back to the initial point.

[kunal]

@Matt J Here I am done with filtering the cluster points

[Image Analyst]

Some questions:

1. Is there some reason why you want to connect the inside corner points to the outside corner points? The diagonal outside corner point would be one of the farthest nearby points on the other side of the thick line. If you chose the closest point, the line would be shorter because it would go straight across. Is there some reason you want to make it longer and go farther? Or not really - it's just because of aesthetics reasons?
2. How were the red points determined in the first place?
3. What will you do with the information once you have the red line segments determined? Are you going to split apart the thick white line into separate segments? If so, then what are you going to do?

https://en.wikipedia.org/wiki/XY_problem

[kunal]

The red points are the corner points where there is a drastic change in the angles and i am later using using connected point to divide this mask area in different 4 point segments that will be similar to rectangle.

[Image Analyst]

I don't think this addressed any of the numbered questions. I don't see how that polygon would be divided into 4 rectangles. Anyway, if you go to the corners, the tips of the thick lines will be pointy arrowheads, not rectangles. But let's say you want to split that thick white line/polygon into 4 or more smaller rectangles. Why? And then what are you going to do with those rectangular masks?

I'm just trying to help but want to make sure you don't go on a wild goose chase for some solution you think will work but might not work.

https://en.wikipedia.org/wiki/XY_problem

Answer 1

[Matt J]

load('coping_data (1).mat')

img=logical(img);
n=height(points);

f=@(z,n) circshift(z,-n);
mod1=@(z) mod(z-1,n) + 1;

x=points(:,1)'; dx=diff([x,x(1)]);
y=points(:,2)'; dy=diff([y,y(1)]);

i=find(dy<-5 & f(dx,1)<-10 & f(dy,2)>+5 & abs(f(dx,1))<50 ,1);

I=mod1( i+2:i+n/2+1 );
J=flip( mod1( I(end)+(1:n/2) ) );

X=[x(I);x(J)]; Y=[y(I);y(J)];

imshow(img,[])
line(X,Y,'Color','r','LineWidth',3)

Answer 2

[Image Analyst]

If you can get the inner boundary and outer boundary as separate arrays, then it would be easy to find the boundary point closest to the red points by using pdist2. If you can't figure out pdist2 then I can do it if you give me the inside boundary and outside boundary.

[kunal]

Ok. Let me try it