Research on Image Super-Resolution and Detail Preservation Technique Based on Non-Local Information

In this paper, a method for constructing an image super-resolution model based on nonlocal information is proposed, which covers the key steps of the resolution process, nonlocal information extraction and feature selection, image similarity block matching, and image reconstruction, so as to realize the super-resolution processing of images. In addition, a set of image detail preservation strategies are designed, which include image information preservation and a balanced mechanism of detail preservation and noise suppression to ensure that the detail clarity of the image is maintained during the super-resolution process. After validation, the perceptual loss of this method reaches 4.8083, which is a remarkable achievement especially in medical images, and the FSIM index is as high as 90%. This research not only effectively preserves the detail information while improving the image resolution, but also provides a valuable reference basis for future in-depth exploration in the field of image processing.