From the view point of the parameter estimation, the model of the least squares matchgin (LSM) can be taken as an adjustment model with additional parameters. Thus the problems which exist in the LSM could be analyzed and solved by the theory of additional pa-rameter adjustment. Based on the theoritical researchs and experimental analysis, the additional parameter is one of the most important factors to influence the qualities of LSM. In this pa-per, a new algorithm called PDFM is proposed,which the parameters(P) dynamic (D) filtering (F) technique is incorporated in the LSM. The obtained experiment results shows, that the PDFM algorithm improves the qualities of LSM at accuracy,speed,iteration times and pull-in range.