Low-Light-Level Image Super-Resolution Reconstruction Based on a Multi-Scale Features Extraction Network

Wide field-of-view (FOV) and high-resolution (HR) imaging are essential to many applications where high-content image acquisition is necessary. However, due the insufficient spatial sampling of detector trade-off between pixel size photosensitivity, ability current sensors obtain high resolution limited, especially under low-light-level (LLL) conditions. To solve these problems, we propose a multi-scale feature extraction (MSFE) network realize pixel-super-resolved LLL imaging. In order perform data fusion information for low (LR) images, extracts high-frequency detail from different dimensions by combining channel attention mechanism module skip connection module. this way, calculation components can receive greater attention. Compared with other networks, peak signal-to-noise ratio reconstructed was increased 1.67 dB. Extensions MSFE investigated scene-based color mapping gray image. Most could be recovered, similarity real reached 0.728. The qualitative quantitative experimental results show that proposed method achieved superior performance in fidelity enhancement over state-of-the-art.