According to the large scale spatial objects real-time attitude measurement demand in the process of large equipment assembly and docking in the field of aviation and aerospace, a new method for attitude measurement is put forward based on orthogonal cylindrical imaging, disassembling and reconstructing the function of two-dimensional angle measurement which is provided by the existing array image sensor using in photogrammetric measurement system. The method makes full use of the advantages such as high one-dimensional angular resolution, fast acquisition speed and simple treatment of linear array CCD, and simplifies the camera structure with cylindrical orthogonal imaging optical path. A non-parametric calibration method is used to correct the aberration of the imaging cylindrical. In order to solve the coordinate synchronization problem during space object attitude measurement, the real-time recognition tracking method is researched based on extended Kalman prediction to realize parallel measurement of multiple targets, and then realizes the real-time attitude determination through attitude calculation model based on Rodrigues parameters. Experimental results show that space three-dimensional coordinate measurement accuracy is better than 0.5 mm, and the maximum measurement errors of space object pose solution in yaw, roll and pitch directions are 0.20°, 0.12° and 0.23° respectively. The proposed method has high accuracy of attitude measurement.