Image Super Resolution Reconstruction Based on Multi-Level Residual Skip Connection Network

Image super-resolution reconstruction technology can convert low-resolution images into high-resolution images with higher pixel density and clearer details, and plays an important role in military and medical fields. Aiming at the problem of insufficient processing of texture details and color restoration degree in existing image super-resolution reconstruction algorithms, a multi-level residual skip connection network (MRSCN) based on coordinate attention mechanism is proposed and applied to the SRGAN model to realize full utilization of low-resolution image features. It is used to recover the details of the image and to optimize the perceived loss using Charbonnier loss and TV loss. This algorithm is tested on Set5, Set14, Bsd100 and Urban100 data sets for 4x super-resolution reconstruction. Compared with other commonly used super-resolution algorithms, this algorithm can better retain texture details during image reconstruction, resulting in clearer image details, better visual effects and effective reduction of the number of parameters in the network. In terms of objective evaluation indicators, the average value of PSNR and SSIM increased by 0. 503dB and 0. 0076 respectively compared with the original SRGAN.