How can I find the skeleton of a human body ?

Hello everyone !

I am doing my summer internship in the reserch department of the medical field and I have some troubles with my project. I work with image processing and I need to find the "Ideal" skeleton of a human from a binary image (as a 5 years old can draw it). I already used the bwmorph() function but as you may know the result looks like a subway map, and I only need the "principal lines" of the body (legs, arms, torso and head) without the littles excrescenses.

I found some topics on the internet about this but it doesn't work very well...

If anyone can help I'll be grateful !

Thank you in advance ! (and sorry for my bad English)

Answer 1

[Image Analyst]

You can look up papers on it here: http://www.visionbib.com/bibliography/contents.html. Use the search function to look for words like stick figure, gait, gesture, etc.

You might want to take a look at Cleve's blog on the eigenwalker : http://blogs.mathworks.com/cleve/2016/04/11/the-eigenwalker-model-of-the-human-gait/?s_tid=srchtitle

[Paul BLONDE]

Hello and thank you for you reply. I've found many papers that could have an answer but they're all for sale... Do you think it is possible to obtain the joints of a stick figure with only a 2D binary image of a human body ? I know how to connect the joints using the line() function but I can't figure out how to detect the body parts and "put" the joints on those parts.

I use a thermal camera coupled with a RGB camera (visible light) but I don't have any depth informations (unlike Microsoft Kinect).

Thank you in advance for you answers.

[vps]

Hi Paul, Can you please tell us that which algorithm or research paper did you use? Thank You

Answer 2

[saranya kanthaswami]

Would you please give the code to obtain the skeleton

Answer 3

[NURUL AMIRA MHD RIZAL]

Hai. May I know how to find a human body skeleton in an image? I am using a webcam to capture the image.

[Jan]

Please do not attach a question in the section for answers of another question, but open a new thread. Afterwards delet this message - thanks.