How to smooth binary image

I have the binary image below which I want to use Hough Transform to detect the line in the middle. However, it cannot detect it. How can I smooth the image so the line is more easily detectable for the hough function? Thank you

Answer 1

[Matt J]

load BWImage; 

BW=(1-entropyfilt(BW,ones([11,5])));

BW=imclose(BW,ones([11,1]));

BW=bwareaopen(BW>0.2,20);

imshow(BW,[])

[ijsonvjksrefdsb]

It worked well with the hough function, thank you!

[Image Analyst]

But why do you think you still need the hough function after this? If you want you can just use aspect ratio or bwareafilt to extract the tall, narrow blob. What are you going to do with it, once you have the blob of interest isolated?

[ijsonvjksrefdsb]

@Image Analyst I'm sorry for the late reply, I'm trying to recreate a research paper on the automatic detection of solar radio bursts which uses radon transformation.

[Image Analyst]

OK, thanks for the context. But still, why do you need Hough? Most of the blobs are line-like so, depending on the sensitivity hough may find them all, or find none, or find blobs other than the one you want. I don't know what the paper did but I think the problem in the binary image -- almost everything is connected leading me to think that there were better ways to segment the image than whatever you did.

[Matt J]

The morphological operations in my answer may not always seal the breaks in the line of the interest. The FillGap option of houghlines would probably be a good way to hedge against this, as opposed to purely region-based analysis.

[ijsonvjksrefdsb]

@Image Analyst They used radon on the binary images to get the max and min values for a background subtraction, I attached the paper here, I only followed what they instructed in pages 3-4 for the binary images, then about radon in page 5.

[ijsonvjksrefdsb]

@Matt J I see, I'll check it out, thanks a lot!