There are a few possibilities.
You could use pattern matching to track the pendulum. If you included the arm, you could get the rotation angle from the pattern matching. Plotting the angle vs. time would give you the period, or you could use frequency analysis to get an exact value.
You could draw a line ROI horizontally that the pendulum always intersects. Searching for the peak or valley along the ROI will give you the position. Once again, you could plot it or use frequency analysis (FFT) to get the period.
You could just look at a single point that the pendulum passes through. It will pass through twice per period, so you could measure the time between every other pulse. For this, you would need to extract the point from each image in a sequence and plot the intensity vs. time. Frequency analysis will work here also. This method is a little trickier, because you have to use the maximum shutter time possible for the frame rate. If you don't, you could miss the passing of the pendulum.
I'm sure there are a few other ways to do it as well. If none of these will work, explain the problem with a little more detail.
Bruce
Bruce Ammons
Ammons Engineering
