Payload attitude is inevitably affected by installation error factors such as platform jitter

and sensor thermal deformation error

to some extent

when satellite is in orbiting

that leads to lower precision of image localization. A real-time attitude calibration algorithm for space-based optical sensor based on feature points is presented. Firstly

the state model of the installation error and observation model based on feature points are established. Secondly

an EKF is used to achieve real-time estimation of installation error. Simulation results show that the algorithm can achieve real-time attitude calibration and has stable performance

and can effectively improve the precision of image localization.