For the star observation system

because of the movement of CCD flat

the CCD star serial images exist global movement. So

it is needed to estimate the motion parameter precisely. The invention and development of particle image velocimetry(PIV) indicate the important breakthrough of modern experimental fluid measurement. PIV is important both in the theoretical studies and application of the fields of fluid mechanics and aerodynamics. A motion estimation algorithm of serial Star images based on self-adaptive window selection PIV technique is proposed in this paper. The algorithm chooses analysis window of maximum energy by a self-adaptive method. Then the global motion parameter can be estimated through cross-correlation of the two analysis window. The experimental results demonstrate the algorithm is of high accuracy

and reduce the estimation error to be less than 0.01 pixel.