Given the material characteristics and manufacturing process of refrigerated infrared detectors, pixel inconsistency is common in infrared images. This non-uniformity of infrared images can significantly degrade the quality of on-orbit imaging. In particular, detecting point targets against a uniform background becomes more challenging, often increasing the false alarm rate during on-orbit detection. The two-point correction method is a typical processing method for addressing image in-homogeneity in infrared remote senor data in orbit. This method can effectively restrain image in-homogeneity to a certain extent. However, the correction effect of this image processing method only depends on imaging parameters of the camera but also on the working temperature of the detector in the time domain. Therefore, even if the imaging parameters of the optical remote sensor of the refrigerated infrared detector are fixed in each imaging mission, the correction coefficients must still be calculated in real time. This is achieved by obtaining internal calibration images of the on-board calibrator, thereby ensuring that the heterogeneity of the infrared image is suppressed. Based on a linear push-sweep mid-wave infrared remote sensing camera, an adaptive time-domain correction algorithm in is proposed. Currently, the algorithm has already been deployed in on-orbit optical remote sensors, where it effectively reduces the frequency of in-orbit calibrations and enhances the processing capability of emergency imaging in the infrared exterior passage way.