Abstract: In the video image acquisition process for high speed moving object,the video and image shake due to high speed,wind and other factors．In order to improve the high speed moving target image acquisition performance of vision system,and improve the image quality,an improved algorithm of image stabilization for high speed moving target images based on adjacent frames compensation was proposed． We preprocessed the image by combining the gray histogram equalization method and adaptive filter, extracted the feature points in the video image through the scale invariant feature transform(SIFT) algorithm．And the affine model was used to solve the motion parameters．Kalman filter was used to scan the video image．Finally we used the adjacent frame compensation methods of a previous frame image as the reference frame of parameter compensation to the current frame,and the electronic image stabilization processing was realized．Simulation results show that the new algorithm can retain the features in the image while removing images shake and jitter,and is very suitable for electronic stabilization processing for high speed video image．The precision is improved and the computing consumption is reduced significantly.

Key words: High speed moving target,Image,Image stabilization algorithm,Frame compensation