The similarity transformation model between different coordinate systems is not accurate enough to describe the discrepancy of them.Therefore,the coordinate transformation from the coordinate frame with poor accuracy to that with high accuracy cannot guarantee a high precision of transformation.In this paper,a combined method of similarity transformation and regressive approximating is presented.The local error accumulation and distortion are taken into consideration and the precision of coordinate system is improved by using the recommended method The similarity transformation model between different coordinate systems is not accurate enough to describe the discrepancy of them. Herefore, the coordinate transformation from the coordinate frame with poor accuracy to that with high accuracy can not guarantee a high precision of transformation. In this paper, a combined method of similarity transformation and regressive approximating is presented. The local error accumulation and distortion are taken into consideration and the precision of coordinate system is improved by using the recommended method