High dynamic range imaging technology can reflect scene information comprehensively and effectively, which is beneficial for obtaining higher imaging qualities in high dynamic range scenes. However, the classic high dynamic range image fusion method of using a single camera through multiple-exposure fusion tends to result in the “ghost” problem in a dynamic scene, whereas the method of using multiple sensors in a simultaneous exposure system is complicated and expensive. Meanwhile, an extension method based on a single low dynamic range image loses details easily in underexposed or overexposed areas. These methods are often used under better lighting conditions. Hence, a high dynamic range image fusion method based on a dual-channel low-light-level (L3) CMOS camera is proposed for low illumination dynamic scenes. First, an image acquisition platform built using a dual-channel L3 CMOS camera is used to collect two images with different exposures for low illumination dynamic scenes. Based on the accumulative histogram, the principle of dynamic range extension is established, and the two images collected by the system are extended. Finally, the pixel level fusion method is used to fuse the sequence images after the dynamic range extension. The results show that the method of dynamic range extension fusion can yield high dynamic range images under L3 dynamic scenes as well as better imaging quality.