Circular synthetic aperture radar (CSAR) has the capability of digital elevation model (DEM) extraction. A DEM extraction method using CSAR subaperture image sequences is presented in this paper. In order to increase the precision of the extracted DEM, the correlations among all the subaperture images in correlation within an arc are utilized in this method. The joint cross-correlation coefficient of all the subaperture images in correlation within an arc is taken as a measure function, which has a higher sensitivity in height extraction. Meanwhile, the geometrical distortion in the subaperture images is corrected by projecting them on 3D space. Their correlations are thus improved. The results and analyses of the gotcha public release dataset validate the effectiveness and accuracy of the proposed method.