Feature detection of a profile view picture?

Hi,

I am a student currently working on a project where we want to identify the bridge of the nose and the end of the chin from a profile view picture automatically. I was thinking that it may be possible to identify the bridge of the nose using Matlab's edge function to find a trace of the outline of the face and then trying to find a local minimum where the bridge of the nose would be. Are there any cleaner ways to find these points that anyone knows of?

Thanks for any ideas!

[Jan]

Sorry for this unuseful reply, I do not post it as an answer intentionally: Why bothering Matlab with techniques, which cannot be 100% reliable, when a 2 year old child can solve the problem in fractions of a second? So buy 8 touch screens (multi-threading!), go to a kindergarden and offer cookies for pointing with the finger to the specified regions.

This is not meant as a joke and it is known, that I appreciate Matlab for its efficiency and power. But sometimes using Matlab to solve specific problems is like using the handle of a drilling machine to push a nail into the wall.

[Image Analyst]

(1) Why do you want to know the location of this point? (2) Where did you upload your image?

[Anand]

You need to provide more details than that for a real answer.

If you can't already find the profile face, take a look at the vision.CascadeObjectDetector in the Computer Vision System Toolbox. There is a classification model for a profile face that you can use. There's also a model for nose and eyes, but I think they only work for a frontal face, but I can't be certain.

Once you have the profile face, use the edge function to get the contour of the profile face. You'll obviously have to adjust the parameters to get this right.

Post the image you get after that and then we can move ahead.