For larger light collection area and detection sensitivity, the aperture of future telescopes is becoming larger and larger. The large sparse aperture telescope is one of the important tools for the future astronomy. Currently, the largest sparse aperture telescope is the Large Magellan Telescope (24 meters). For large sparse aperture telescopes, it is necessary to focus the direction of multiple sub mirrors on a single point, namely, co-focus. Co-focus (optical axis alignment) for sparse aperture is the foundation for achieving the established scientific goals of telescopes, and it is also the key to achieving functional synthesis of multi-channel systems. The dynamic range of wavefront sensing required for relatively large co-focus detection. For Hartmann sensors, a large sub aperture can be used to enhance the dynamic range, but it can lead to the addition of other aberration components in the sub aperture, deviating from the Gaussian assumption. For application scenarios such as fine astronomical observation, there will be huge limitations. At the same time, an excessive dynamic range can exacerbate the nonlinearity of the solution, leading to control degradation of the wavefront correction system.The co-focus measurement of large sparse aperture telescopes is mainly divided into two methods of using the focal plane and the pupil plane. Here, we first obtain two defocused star point images before and after focusing, and calculate the wavefront based on the wavefront curvature sensing method through the defocused star donut image. Afterwards, a mask is used to suppress the edge noise of the star image, which involves physical constraining. Then, co-focus perception and regulation can be achieved through plane fitting. The co-focus detection method for sparse aperture telescopes based on curvature sensing is shown (Fig.2), which overcomes the disadvantage of traditional methods where light points overlap and are lost, requiring recalibration.By using segmented deformable mirrors to generate higher-order wavefront aberrations, the ability of curvature sensing to perceive boundary anomalies can be verified through testing the wavefront tilt. The verification platform for wavefront tilt calculation using deformable mirrors is shown (Fig.6). By utilizing the tilt component obtained from wavefront sensing, combined with the mapping relationship between aberration space and mirror space, the driving force of hard points can be obtained through inverse solution. In this analysis, it is assumed that the system has a total of 7 primary mirrors, each with tilt errors. By utilizing the difference in light intensity before and after focus plan, wavefront sensing is achieved, and closed-loop correction is achieved based on this. According to Fig.3 and Fig.4, the total adjustment range is greater than 20 wavelengths.By utilizing the non-interference, wide range, and band robustness characteristics of curvature sensing, co-focus measurement of large sparse aperture telescopes is achieved. This method can achieve parallel perception and control of multiple mirror tilts without performing focus recognition. In this analysis, it is assumed that the system has a total of 7 primary mirrors, and each sub mirror has tilt errors, utilizing the difference in light intensity before and after focusing. Finally, the correlation between the wavefront tilt detection results is higher than 0.83, and the measurement accuracy is better than 0.2λ （λ=633 nm）. By utilizing curvature sensing, the drawbacks of multiple moving calibrations and individual adjustments can be overcome. High throughput co-focus error perception and rapid regulation are realized. The method in this paper lays a technical foundation for the construction of future large aperture sparse aperture telescopes.