针对目前CMOS探测器的噪声研究绝大部分采取频率域的分析方法, 所得的结果对于实际的探测器应用指导性不强。从时间域出发, 构建了全新的探测器噪声模型, 分析了在探测器的不同工作时期内噪声量大小同时间的关系。给出了探测器工作时序的优化准则, 积分时间越长, 系统的信噪比增加, 输出信号电压值增大, 等效输入端信噪比不变, 系统的动态范围减小; 复位时间越短, 输出噪声量越小, 但是图像的滞后效应也就越严重, 在对弱小目标进行成像时应该尽可能地减少复位时间, 增大积分时间, 对于亮目标成像, 应该增大复位时间, 减少积分时间。

There are a lot of researches have done on CMOS detectors, but most are taken by the frequency domain analysis which is not useful for the actual detector applications. This paper describes a new noise analysis way combined with specific application examples in the time domain, which gives us a new output noise formula and a new expression of signal to noise ratio. At last some suggestions have been made of how to get better applications on CMOS detector. The conclusion is that, when the integration time is longer, the SNR of the system will be greater, the output signal voltage will increase, but the equivalent input signal to noise ratio is unchanged, so the dynamic range of the system is decreases. And when choosing a shorter reset time, the amount of output noise will be smaller, but the image lag effect is more serious, so when imaging of weak target we should reduce the reset time, increasing the integration time; on the other hand for bright target imaging, we should increase the reset time, reducing the integration time.