Aiming at the problems of low illuminance in the cabin and dark and poor-quality images captured by the on-board camera in the ejection seat ejection test, this study proposes a low-light image enhancement method based on the improved U-Net++ . The nested jump connection has been optimized to enhance the understanding of local and global information by U-Net++ , making it more stable in low-light scenarios. Introduce U-Net ++ as an additional computing node between the encoder and the decoder to enhance its perception ability of low-light regions and improve the quality of the image enhanced by this algorithm; Reduce the detail flow in the U-Net++ structure and retain the details in the dark areas; Aiming at the problem of the missing reasoning part of the relationship between pixels in U-Net++ images, a post-processing procedure based on Transformter enhancement was designed to solve the problems of halo and inconsistent light intensity in the processed images. The performance of the algorithm in this study was analyzed. The algorithm was applied to the public low-light image dataset and the actual captured low-light images. By comparing the image processing quality of the algorithm in this paper with that of the existing algorithms, the superiority of the low-light image enhancement algorithm based on the improved U-Net++ was verified.