Lightweight image deblurring method based on improved Transform

The invention discloses a lightweight image deblurring method based on an improved Transform. The lightweight image deblurring method comprises the following steps: acquiring a blurred image; inputting the blurred image into a trained lightweight image deblurring model based on the improved Transform to obtain a clear deblurred image corresponding to the blurred image; wherein the lightweight image deblurring model based on the improved Transform comprises a coding module, an intermediate module and a decoding module, the coding module is composed of a dynamic convolution residual module, and the intermediate module and the decoding module are composed of a simple Transform module; due to the fact that the simple Transform module can obtain global information and position information in space and only needs calculation complexity close to linearity, the lightweight image deblurring method based on the improved Transform can obtain high-quality deblurred images in a very short time. 本发明公开了一种基于改进Transformer的轻量级