Here is my homework 5.

For the old movie, the use of two opens and two closes seem to give the best 
results; almost all noise is eliminated, and the signal is retained, 
although certain shadow effects still occur, distorting the image from the 
human form. However, it would not be possible to determine shadows from 
people given one camera in only black and white. However, the signal/noise 
ratio is much improved over the previous version.

For the cars, FGclag seems to give the best results. The reason for this is 
the camera jitter. Because the camera moves throughout the capture, the 
average of the lowest histograms does not include the road background for 
the higher histograms, which are in a different place. However, averaged 
over the entire movie, the road markings and the parked cars are treated as 
background for almost the entire time.Certain cars may be more difficult to 
see, but there are definitely moving strucutres that by and large keep their 
shape. In addition, the vehicles are the only moving objects detected; even 
the noise particles tend to stay in almost the same position, whereas the 
vehicles go completely across the field of vision.

Before running the cars file, load the movie and run the prelim file to 
generate the equalized images, which seem to work better than power law 
transforms.

Thomas Young