This paper proposes a lightweight low-illumination image enhancement network inspired by the Retinex theory. In the model proposed, a pixel-wise adjustment function realizes the lightweight of the network structure; the optimization bottleneck problem is solved by introducing the shortcut mechanism. Through the test, the SSIM index of the proposed model is 7.04% higher than that of MSRCR, and 31.03% higher than that of CLAHE. In the actual use case, the proposed model can process videos with a resolution of 400×600 at a speed of 20fps on average, which meets the requirements of DMS video stream processing speed. Also, a MobileNet distraction state recognition network pre-trained on the SFD dataset is used as the back-end to verify its application in the DMS system. The results show that the recognition accuracy of the driver's distracted behavior in a low-light environment is improved by 75.39% compared to before use.

To run the program, PLZ download from the master branch!!!