X-ray single photon detection system convert the x-ray single photon to electrical signal for x-ray pulsar detection, etc., which commonly use detectors as SI-APD (Silicon - avalanche photodiode), SDD (Silicon drift detector), Si-PIN (Silicon PIN photodiode). The x-ray pulsar detection requires higher time-accuracy sensor corresponding to only one pixel. The energy of x-ray pulsar is very weak, approximately 2.3 counts/s•cm2 in LEO(Low earth orbit). In order to improve the efficiency of detection, detection array of multiple detectors is used to increase the effective detection area. The project of this paper selected 8×10 SDD sensor array. Each detector has independent readout circuit. Therefore, system gain and response time are different, resulting in nonuniform response to homologous target, this nonuniformity includes the gain response nonuniformity and the time response nonuniformity, in this paper, the gain response nonuniformity correction method is discussed. The usual method of gain response nonuniformity correction is using the signal source as the circuit input instead of the detector, which has inevitable consequence: the signal source must have extremely high precision and the signal Integrity, such as cable interface or connection distance, is indeed quite sensitive because of the huge gain of the single photon detector circuit, which is usually thousand-fold. Moreover, the usual calibration method can’t correct the detector's own response nonuniformity, and the efficiency is low through the one-by-one correction. In this paper, the "characteristic peak correction method for specific target materials" is proposed, and the implementation of the method is as follows: bombard the characteristic target using high-energy x-ray source (>40keV), get the target material characteristic peak by calculating the amplitude histogram of the output signal (characterization of x-ray Single photon energy), and obtain the correction coefficient after normalization. For example, the Fe Target has main characteristic peak of Kα 6.494keV, Kβ 7.058keV as well as impurities characteristic peak of 3.692keV (Ca), 2.014keV(P), 0.277keV(C). The correction coefficient can be obtained by using least squares method. The correction method calibrates the nonuniformity of 80 detectors and obtains the correction parameters of different detectors at the same time. The experimental results show that the correction accuracy of the characteristic peak correction method for specific targets is better than 0.5%, which is much higher than the traditional method of signal source correction. The correction system for this paper uses FPGA+MRAM architecture, where the MRAM is used to store correction parameters, and to achieve real-time corrections in orbit. The Simulink-SG Modular design language is used. The MRAM is especially used to save the correction parameters, so that uploading correction parameters in orbit is available. Through the specific target detection in orbit and the ‘characteristic peak correction method for specific target materials ’, the correction can be finished in orbit, that ensures the calibration accuracy of the X-ray pulsar detection system at the end of life.