IMAGE MOTION ESTIMATION FROM MOTION SMEAR - A NEW COMPUTATIONAL MODEL

Motion smear is an important visual cue for motion perception by the h uman vision system (HVS). However, in image analysis research, exploit ing motion smear has been largely ignored. Rather, motion smear is usu ally considered as a degradation of images that needs to be removed. I n this paper, we establish a computational model that estimates image motion from motion smear information-''motion from smear.'' In many re al situations, the shutter of the sensing camera must be kept open lon g enough to produce images of adequate signal-to-noise ratio (SNR), re sulting in significant motion smear in images. We present a new motion blur model and an algorithm that enables unique estimation of image m otion. A prototype sensor system that exploits the new motion blur mod el has been built to acquire data for ''motion-from-smear.'' Experimen tal results on images with both simulated smear and real smear, using our ''motion-from-smear'' algorithm as well as a conventional motion e stimation technique, are provided. We also show that temporal aliasing does not affect ''motion-from-smear'' to the same degree as it does a lgorithms that use displacement as a cue. ''Motion-from-smear'' provid es an additional tool for motion estimation and effectively complement s the existing techniques when apparent motion smear is present.