A motion estimation algorithm of serial star images based on adaptive window selection PIV technique is proposed.The algorithm chooses the analysis window of maximum energy by an adaptive method.Then,the global motion parameters can be estimated through the cross-correlation of two analysis windows.The experimental results show that the presented algorithm improves the precision of parameter estimation,and overcomes the influence introduced by stars moving in or moving out of the analysis window.The algorithm reduces the estimation error to less than 0.01 pixel and can estimate the global motion parameters with high accuracy and high speed.It provides a good way for solving star image matching.