Phase fringe image speckle noise suppression method based on deep learning

The invention discloses a phase fringe image speckle noise suppression method based on deep learning. The method comprises the following steps: S1, collecting a phase fringe image data set; s2, adding speckle noise with random noise levels to the training set data samples; s3, preprocessing the training set data added with the noise; s4, sending the preprocessed sample into an asymmetric fusion non-local block structure; s5, inputting a result obtained by processing in the step S4 into an FFDNet network for noise suppression; s6, sending a denoising result obtained in the step S5 to an edge extraction module; s7, building a network model, and training the network; s8, speckle noise of different noise levels is added to the test set data sample according to the step S2; and S9, sending the test set sample obtained in the step S8 to the network model trained in the step S7. According to the deep learning method, under the condition that network parameters are not changed, speckle noise of ...