Abstract: Luminous remote sensing images have been widely used in social and economic estimation, urban monitoring, ecological environment assessment and public health. Using luminous remote sensing to detect ground objects in low illumination conditions such as nighttime and twilight, it is complementary to traditional daytime remote sensing to form all-day earth observation capability. However, in low illumination environment, the image quality of remote sensing camera will decline sharply with the attenuation of camera light input, so how to improve the signal-to-noise ratio (SNR) of remote sensing image is the focus of luminous remote sensing. The SNR of a remote sensing camera developed in the early stage was analyzed and modeled in detail. The functional relationship between SNR, integration time and ground radiation brightness was given in the face array staring mode. When the integration time is greater than 2 ms, the SNR is better than 10 dB, and when the integration time is greater than 15 ms, the SNR is better than 20 dB. In strip imaging mode, the relation between SNR and line frequency and TDI series was given. The results show that when the TDI series is constant, the SNR decreases with the increase of line frequency, and when the line frequency is constant, the SNR increases with the increase of TDI series, but the speed of SNR increase gradually slows down with the increase of TDI series.