Research on Application of Low Illumination Image Enhancement in Energy System Monitoring Based on Improved Zero-DCE Model

The problem of poor low-light image quality is common in nighttime or complex environment monitoring of energy systems, which seriously affects the readability of images and subsequent analysis and processing. This paper proposes a low-light image enhancement method based on the improved Zero-DCE model. It uses a lightweight convolutional neural network structure to achieve adaptive enhancement of image brightness and details without needing reference images. Experiments on the LOL dataset and energy system monitoring images show that the model outperforms mainstream methods on multiple image quality indicators and has good visual effects and engineering application value. The model can be embedded in existing camera systems or edge computing devices to achieve real-time enhancement of lowlight images, providing clearer visual input for tasks such as remote monitoring, manual inspection, and fault identification of energy systems.