Thread Subject: detect and classify movements in a sequence

Dear sirs,
I've little experience in matlab and signal processing in general.
My goal is to classify different type of movements in a video sequence
and then detect areas that moved in the same way, for parking and traffic recognition

I've a tiff sequence in a m X n X Q array, I started to approach
the problem in time domain storing the value of each pixel during
the sequence in a? 1 X Q array and plotting the resulting signal.

for a car that pass fast in the scene, the corresponding pixels signals has peaks.
for a car that stops for a while, the corresponding pixels signals have a square wave shape.??? ?

I want first to distinguish pixels signals that moved in square wave shape from those
that moved in a peaks shape (stationary cars from fast passing cars).
Then I would like to classify the square shape signals that are similar in shape and
correspond to neighbours pixels of the image and assume that this area correspond to a car.

At the moment I've stored all the signals and I don't know how to classify signals in
relation to their shape.

I think this analysis in time domain will avoid complex operation
of pattern matching on cars, to allow almost real time performance.

Any suggestions or guidelines will be kindly appreciated

Hi sirs,
please give me some advice on how to approach this problem.
Even documentation, or readings are welcome,

many thanks