Click here to Skip to main content
15,900,907 members
Please Sign up or sign in to vote.
0.00/5 (No votes)
See more:
I have started working on a visual identification system to identify a person and extract features from a captured video.
For the initial step, I require to segregate and isolate the targeted subject from the image(frame) background and create a binary silhouette(outline) of the person.
The background can be chosen as I like, which would reduce noise to some extend.
I tried to apply canny algorithm in OpenCV but the outline achieved is not continuous and if I apply a floodfill to the within the boundary a silhouette is not achieved.
Can anyone suggest a method in OpenCV using which I may achieve an approximate outline of the person?
Posted

1 solution

Simple outline detection will not be sufficient. You need to use classifier techniques. There are many and I will point you to HOG. Identification of individuals is another problem.

A paper on the theory of HOG detection is here and even if you cannot understand it all will give you some ideas.
http://lear.inrialpes.fr/people/triggs/pubs/Dalal-cvpr05.pdf[^]

This is a good tutorial to begin with:
http://chrisjmccormick.wordpress.com/2013/05/09/hog-person-detector-tutorial/[^]

OpenCV ref:
http://docs.opencv.org/modules/gpu/doc/object_detection.html[^]
 
Share this answer
 
v4
Comments
nv3 30-Mar-14 5:03am    
Good advice. 5.
suramrit 30-Mar-14 5:15am    
Thank you

This content, along with any associated source code and files, is licensed under The Code Project Open License (CPOL)



CodeProject, 20 Bay Street, 11th Floor Toronto, Ontario, Canada M5J 2N8 +1 (416) 849-8900