A mobile mapping system (MMS) which equipped with digital cameras is capable of capturing overlapped image sequences with known image orientations referring to the ground coordinate system. The 3D coordinates of a feature point can be determined through an intersection of overlapped image sequences. This calculation needs the input of conjugate image points of the target feature. This paper proposes an automatic method of conjugate point registration for MMS image sequences. The key concept of the proposed method is image matching with the constraint of epipolar geometry. Based on one image point given by an operator, the algorithm is designed to trace conjugate point along the epipolar line of each overlapped images. The employed image matching methods is cross correlation. The matching quality is evaluated using the correlation coefficient of the multiple conjugate image patches finalized. The proposed method was tested with some image sequences of a real MMS case. The results show the feasibility of the method. This algorithm can largely reduce the manual workload of image measurement for MMS image processing.