How to find the desired element in an image?

Hi, I have an image shown and I want to find the linear structure elements with 80 pixels of length and angle= 30 degree which are fit in the attached image.

How can I do so? Thank you.

[Walter Roberson]

it is not obvious to me that such elements are present at all??

[ZWY]

Hi Walter, actually I want to extract the main arcade of the vessel. The solution I think of is the vessel has the information like certain length and angle, which can help me to extract the main arcade out by using the length and angle information.

[Walter Roberson]

I do not recognize the term "arcade" in this context?

[ZWY]

Maybe can be replaced with main vessel

[Walter Roberson]

bwmorph 'skel' and remove spurs?

[DGM]

In the prior example using bwskel(), I showed the use of the 'minbranchlength' option to selectively prune short branches. That may be one thing you could work with. Selecting branches by angle might be a little more complicated, and perhaps might require some sort of decision regarding how to define the "angle" of a curved line.

The fact that there are so many loops in this image is probably going to be an obstacle though. You might start with a morphological opening to break connectivity in thin areas, but then again, there appear to be some thin spots and breaks in the image of the larger vessels already.

[ZWY]

Hi, actually I have found an article: https://www.sciencedirect.com/science/article/pii/S0010482509002042 .

It is suggested to find the main arcade under the 2.1.1. Detecting the vascular tree. You may refer the part after the equation 5. It uses 24 linear structuring elements, each with 80 pixels of length, designed to fit the main vessels of the arcade in the right and left eyes.

Anyway, thank you for your advice.

Answer 1

[Image Analyst]

It's not worth doing anything with that image because that segmentation is garbage. You need to first improve your segmentation. I believe I told you before so I'm surprised if you took my advice and wrote a paper based on how others did hte segmentation. You can get algorithms here:

http://www.visionbib.com/bibliography/contentsmedical.html#Medical%20Applications,%20CAT,%20MRI,%20Ultrasound,%20Heart%20Models,%20Brain%20Models

Or try using trainable COSFIRE filters for curvilinear structure delineation in images, like this File Exchange entry:

https://www.mathworks.com/matlabcentral/fileexchange/49172-trainable-cosfire-filters-for-curvilinear-structure-delineation-in-images

[ZWY]

Hi, actually I have found an article: https://www.sciencedirect.com/science/article/pii/S0010482509002042 .

It is suggested to find the main arcade under the 2.1.1. Detecting the vascular tree. You may refer the part after the equation 5. It uses 24 linear structuring elements, each with 80 pixels of length, designed to fit the main vessels of the arcade in the right and left eyes.

Anyway, thank you for your advice.

[Image Analyst]

I think I've done that before to enhance fibers. See attached demo.

[ZWY]

Thank you very much