Phase unwrapping method based on global attention upsampling network

The invention aims to provide a phase unwrapping method based on a global attention upsampling network, comprising the following steps: A, constructing a deep neural network structure which specifically comprises a coding network and a decoding network; wherein the coding network comprises five 2 * 2 maximum pooling layers which are connected in sequence, and six PU-M-Net modules; the decoding network comprises five GAU modules, five PU-M-Net modules and a stacking module I; b, constructing a plurality of groups of training sets with different noise levels, and training the deep neural network to obtain an unwrapping network suitable for unwrapping interference graphs with different noise levels; and C, carrying out noise level judgment on the interference pattern to be unwrapped, and then matching the interference pattern to an unwrapping network of a corresponding noise level for unwrapping. The method provided by the invention has relatively high unwrapping efficiency and robustness.