Edge??

friday: Hmm..nice shall celebrate PST's bday watch match and mug for Matrix analysis. Start image processing tomorrow.

saturday: Image processing is easy..will do it tomorrow..Let me do timepass listening to songs and watching stupid movies like "Fitfth element"

late on sunday afternoon: Fuck yeh kya hai? mid sem ke pahele to bahut easy tha..yeh sab kab padhaya usne.. Oh!! shit mein to mid sem ke baad kabhi class hi nahin gaya...

I went into a deep NBD state..and started mugging...by monday afternoon I almost gave up..but somehow kept studying and went to the exam room...I was so nervous that I started laughing in the exam room...Yeah I do that when I am freaked out..How can I ditch this exam? How can I convince the prof to grade me on my mid term? I was thinking all this when the question paper was given to me and it looked very very easy...I started doing the problems quickly...

I had reached as far as the 2nd question when I got stuck and realised I dont know somethings that I need to know..The question was to localise an edge ...forget the technicalities and read on what happened after that...

I walk upto the prof..

Me: Sir, I don't see an edge in this function. Its derivative exists everywhere.

Prof: Oh no! think about the general definition of edge ..

Me: !@#$%^&

Prof: I had discussed it in class...do you remember Marr-Hildreth algorithm!!

Me: Oh yes!! got it. The one with pyramidal decomposition..

Prof: (With a smiling face) Ohh! noooo....Try to remember. It's easy and I have almost(!!!) given you the answer.

(After solving all the other questions xcept one)

Me: I kind of forgot the definition of the unconstrained restoration. Can you jsut remind me. (I dont think I have ever played so much on my luck :))

Prof: Oh what do you think you have to do?

Me: Sir, do "some junk".

Prof : Exactly !! thats write jsut write it down..

Me: thanks..hesitantly..(by now actually embarassed at myself for being so cheap)coming back to the edge problem...I can not remember Mar-Hildreth algorithm.

Prof: Oh..okay..what do you think you have to do to detect edges.

Me: Sir, Are you talking about the thing where we smoothen the image using....hmm... gaussian and then take ...err ..laplacian operatorrr...

Prof: (very excited!!) Exactly!!..(He seemed to be more happy on me being able to crack the funda.) Now since this is one dimensional problem what would be the Laplacian..

Me: (By now completely it ease with a thinking that even if I dont get it he will definitely tell me the solution) Second derivative?? so I jsut take smoothen using Gaussian and then take second derivative?

Prof: Actaully no need to use the Gaussian. That whole thing is another algorithm which uses this method. To localise the edge you just have to take second derivative and then find ..

Me: zero crossing. Okay thanks for your help.

This must have been one of the most funny exams ever taken by me. Anyways the moral of the story is always ask questions if you don't know something, be shameless and never give up ...