随着安防监控和车载摄像对CMOS图像传感器的应用需求越加广泛，微光成像和昼夜兼容成像技术变得越来越重要。拍摄场景中包含的光源、反光物体、阴影等明暗差别很大的目标，需要相机捕捉到可靠的图像信息并能分辨出物体的细节，这对CMOS图像传感器的动态范围提出了更高的要求。 本文设计了一种结合像素内可变电容结构和多次复位技术的动态范围拓展方法，研究了方法原理，分析了关键参数，说明了实现过程。该方法在CMOS图像传感器曝光过程中设置多个复位屏障，防止光电二极管寄生电容提前饱和，并保持信噪比大于能分辨目标细节的最小阈值，拓展了CMOS图像传感器的动态范围。在不同应用环境中通过改变像素内电容大小，同时提高了微光条件下的响应灵敏度和亮光环境下的满阱容量，满足昼夜兼容的应用需求。 本文使用MATLAB进行了模型仿真，试验结果证明该算法能在保持信噪比的同时拓展CMOS图像传感器的动态范围，昼夜兼容成像效果较好。
As the demand for CIS application in security monitoring and automotive camera has widened, low-light-level imaging and day-and-night compatible imaging becomes more important. Recording scenes containing objects with large brightness difference, including light sources, light-reflecting objects and shadows, requires the camera to capture reliable image information and be able to recognize the detail of objects, which needs CIS to have a higher dynamic range. A method of dynamic range widening with in-pixel variable capacitance structure and multiple time reset technique has been proposed in this paper by introducing the principle, analyzing the key parameters and discussing the realization. This method prevents photodiodes from premature saturation by setting multiple reset barrier during exposure process, while maintain S/N ratio being larger than the minimum threshold value of detail recognition requirement, thus expanding the dynamic range of CIS. By changing the amount of in-pixel capacitance according to different application environment, this method increases response sensitivity at low-light-level condition and full-well capacity at strong light environment, satisfying day-and-night compatibility requirements. In this paper, MATLAB model simulation is used and proves that the algorithm expands the dynamic range of CIS while maintaining S/N ratio and day-and-night compatible imaging result is excellent.