The invention discloses a
panorama image stabilization method based on the iteration of global characteristic points. The method of the invention includes two parts of the
estimation of
image motion parameters based on the iteration of characteristic points and the compensation of
image motion based on the wave filtering. The
estimation of global motion parameters includes that a Harries
characteristic point which is the most stable in the image is combined with a characteristic window; local characteristic points are gradually deleted through the identification of the distance consistency and the updating of the iteration of characteristic points so that the global
characteristic point set converges rapidly and the global motion parameters can be accurately obtained. The compensation of the motion image includes that the global motion parameters are filtered to acquire
dither parameters, and actual
dither parameters are compensated and the normal scanning motion of an image
pickup system is retained; simultaneously, the
image fusion is combined with the compensation to realize the output of a
panorama image. The method of the invention can effectively process the dithers such as translation, rotation and zooming, etc., of the image pick-up
system and more complex local motion of the scene, and can be used for removing the influence on the
video sequence by a motion carrier so as to output a stable video and improve the observation effect of the
video monitoring or tracking.