Vibrated image restoration from two consecutive images

Vibrations are often a major cause of image degradation in airborne ad terrestrial reconnaissance, astronomy, robotics, machine vision and computer vision systems. In the case of low-frequency vibration (the exposure time is shorter than the vibration period time) the camera motion during the exposure is random, depending on the vibration parameters (amplitude and frequency) and on the instant of the exposure time. Therefore, the random motion causes a random image blur. Knowledge of the exact motion is practical for calculation of motion optical transfer function (OTF) which can be used with common image restoration algorithms. The motion can be either measured directly using motion sensors or estimated from a sequence of images. Here, a vibration motion estimation method from a minimal sequence of two images is presented. Motion estimation from only two consecutive images is possible due to the use of information hidden in the image blur. The process is carried out in the frequency domain using a new analytical motion OTF calculation method.