Background Subtraction: Various Methods for Different Inputs

With computer vision the problem of speed comes up quite a bit. Images typically only contain a small amount of relevant information which needs to be processed. To help limit the information to be processed a method of pulling out the relevant information is needed. Background subtraction is an often used method to limit the information to be processed, since the background does not contain any "new" information; however, deciding whether an object is in the foreground or background remains a problem. Several methods have been proposed, but there does not exist a perfect method as of yet. Some require extra hardware, require extensive preprocessing, or only perform in certain cases. Two methods of background subtraction looked at in depth include the Gordon, et al. and Wallflower methods. The value of the method is determined not only for its ability to handle the typical problems of background subtraction, but also whether extra hardware is required.