1. Introduction The majority of image sensors used nowadays are composed of CCD and CMOS image sensors [1]. When using these sensors, a simple composition of still images is taken at a predetermined frame rate, to create a video. This approach has several shortcomings; one such problem is data redundancy—each new frame contains information about all pixels, regardless of whether this information is new or not. Handling this unnecessary information wastes memory, power, and bandwidth [2].
