微型电脑应用

在下采样过程中，细微细节和高频信息常因分辨率减小而丢失或变模糊。为了解决此问题，提出图像超分辨率模型，包含浅层特征提取模块、多尺度特征交互模块和重建模块，并对模型进行验证。实验结果显示，所提模型的峰值信噪比和结构相似性指数值分别为31.42和0.862,均方误差和平均绝对误差仅为1.24%和1.39%,明显优于其他方法。

During downsampling, fine details and high-frequency information are often lost or blurred due to reduced resolution. To solve this problem, the research proposes an image super-resolution model, including shallow feature extraction module, multi-scale feature interaction module and reconstruction module, and verifies the model. Experimental results show that the peak signal-to-noise ratio and structural similarity index values of the proposed model are 31.42 and 0.862, respectively, and the mean square error and mean absolute error are only 1.24% and 1.39%, which are significantly better than other methods.