Since there are different applications of radar in various sciences doing research and study on radar is necessary. one of the radar image applications is DEM generation. one of the essential DEM generation steps is image matching. radar image matching is more complex than matching on optical images. one of the most serious problems in SAR image matching is speckle noise. besides the noises, some geometric distortions like shadows, foreshortening, and layover also increase the complexity images matching. in this paper, a SAR image is firstly filtered by a speckle suppression algorithm and then a Harris algorithm is used to extract feature points. Then descriptor is introduced and local matching is performed. Furthermore,we compute the first image point extracted to the second image point extracted displacement with multi quadric equation. Finally, we use least square image matching for image matching improvement. Proposed method for on TerraSAR-X images of the area JAM-IRAN was tested. The result shows that our proposed method is superior to the Local matching in terms of accuracy and number of matched points