How to smooth edges of an image with complex edges?

Hello all,

I am a beginner here. I have been given a project in my college to extract edges of a car door. The edges are to be smooth and accurate. I used edge detection savitky_golay_filter and got an image as follows.

The original image: (reduced the size during upload)

Applied savitky_golay_filter at three different thresholds 40,167,240 and blended all the images to get most of the edges of the above image. After this I took only the red edges and converted the entire image to grey scale.

The below is the resultant:

Now the edges are not smooth and continous. How to smooth the edges so that I get sleek and continous lines?

[KALYAN ACHARJYA]

There is a very popular thread (by Image Analyst) related to the smoothing edges. I didnot get the link right now.

%% Smooth the boundary (by Image Analyst)
windowSize=2; % Change it as per requirements
kernel=ones(windowSize)/windowSize^2;
sm_image=conv2(single(bwImage),kernel,'same');
sm_image=sm_image>0.5; % Change the threshold also, which works for you
figure,imshow(sm_image)

[Harshan Desu]

I tired the same, it is not showing any difference at all. I tired changing the thershold value also. Please help me out.

[Mehmed Saad]

is this smooth enough for you

[Harshan Desu]

Wow that is exactly what I wanted sir, how did u do it?

Answer 1

[Mehmed Saad]

This Comment above

I just changed the threshold condition

red_comp = rbgtest(:,:,1)>rbgtest(:,:,2) & rbgtest(:,:,1)>rbgtest(:,:,3);
windowSize = 10;
kernel = ones(windowSize) / windowSize ^ 2;
blurryImage = conv2(single(red_comp), kernel, 'same');
binaryImage = blurryImage > 0.5; % Rethreshold
figure(2),imshow(binaryImage)

where rbgtest is your blendedoutput.png

[Harshan Desu]

Is there a way to remove the unnecessary edges I want?Can u suggest me how to remove the edges I dont want?