How to Close an Open Contour

Question

Ashkan Javadzadegan 2015-4-18

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/209880-how-to-close-an-open-contour

评论： Image Analyst 2020-6-25

Dear all, Attached image consists of two open contours. I would like to close these two contours. Please let me know if this is doable in Matlab.

Thanks Ashkan

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Image Analyst 2015-4-18

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/209880-how-to-close-an-open-contour#answer_175706

在 MATLAB Online 中打开

burn_multiple_lines_into_image.m

Skeletonize the blob with bwmorph(), then use bwmorph() again to find the endpoints. Use regionprops() to get coordinates of the skeleton so you know which endpoints belong to the same blob. Then use imline() to draw a line connecting the endpoints of each blob. Demo of imline used to burn a line into an image is attached. Please try these steps yourself first and come back with your code if you can't do it. Again, the rough steps are

bwmorph(binaryImage, 'skel', inf);
bwmorph(binaryImage, 'endpoints');
[labeledImage, numBlobs] = bwlabel(binaryImage);
measurements = regionprops(labeledImage, 'PixelList');
for k = 1 : numBlobs
    % find endpoints of this blob, then
    imline();
    % etc
end

Finish it up (it shouldn't be too hard) and let me know of any problems.

5 个评论
显示 3更早的评论隐藏 3更早的评论

Image Analyst 2019-8-9

Didn't I outline the solution : use imline() to join the endpoints of each skeleton. Can't you find the endpoints of each blob? Hint: use ismember() or a for loop to find out which endpoints belong to regionprops(k).PixelList().

darova 2020-6-25

i accepted the answer

请先登录，再进行评论。

Answer 2

Daniel Thomas 2020-6-25

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/209880-how-to-close-an-open-contour#answer_456718

Don't "burn lines" with imline. Use insertShape instead to draw a line on an image. Additional trouble with insertShape is that it always produces RGB instead of grayscale. So you need to convert it to grayscale. Other than that, it works a lot better than "burning the lines."