A Study On Scale Change In Radial Motion

When the camera moves towards the objects, the target images acquired look like be enlarged. If the focus is invariant, then the scale change between sequential frames keeps consistent for the whole image. To estimate the scale change of the image sequence is very important in image registration and electronic image stabilization. This thesis proposes a novel spatial-domain method based on integral gray projection to estimate the scale change in real-time. According to the characteristics of the gray projection curves, a pair of maximum and minimum points are found near the central position of each gray projection curves, the differences in position is calculated respectively and the scale change parameter is obtained as the ratio of both differences. Experimental results show that this method is feasible and efficient.During the exposure period, if there is radial motion between the camera and the target, the image will degrade with radiant blur from the center to the periphery of image, which is the integration of numerous instantaneous exposes of enlarged targets towards the camera or shrunk targets away from the camera. By this analysis, it is pointed out that there is defeat in the radial motion-blurred images directly generated from Photoshop, hence being inappropriate to be used as input deblurring simulation. A pre-processing method is thus given to generate the proper version for simulation.