An algorithm of global motion estimation and compensation applied to electric digital image stabilization was presented.The local motion vector of the block was determined by feature point matching.Globe motion vector between the reference frame and the current frames is obtained by solution of a set of linear equations which come form a given motion model.The parameters to compensate for the current image can be determined by average value filtering and have been applied to give motion model to stabilize the image stream.In order to reduce the computation complexity and speed matching a multi-resolution strategy was adopted.The experiments show that the algorithm has a better effect for improving dynamic image stabilization.