Abstract: Super high resolution (SISR) images are a very hot topic in the field of image processing to convert a single low resolution (LR) image into a super high resolution (SR) image. As part of the real-time imaging function, the softedge reconstruction network is a CNN model that is used to directly reconstruct soft-edge images from LR images. The edge network technology works independently to achieve smooth image edge reconstruction, or subnets are implanted in each SR model to provide smooth front-end image edges to achieve high-quality SR image reconstruction. For super resolution, early researchers used traditional image processing methods. The proposed system is used to improve image unevenness based on the contrast distribution of the object and the edge of the object's detailed description can be found in the picture.

Keywords: Image Dehazing, Contrast Adjustment, Laplacian approach, Deep neural network, High resolution.

| DOI: 10.17148/IJIREEICE.2021.9510