No-reference image quality evaluation method based on countercurrent shape sorting optimization

The invention discloses a no-reference image quality evaluation method based on countercurrent form sorting optimization, and mainly solves the problems that the global dependency relationship of image content cannot be effectively captured and the subjective and objective consistency of image scoring is poor in the prior art. According to the implementation scheme, L RGB images and subjective quality score labels of all the images are obtained from a public website to form a training set and a test set; constructing a reference-free image quality evaluation model comprising a feature extraction module, a multi-axis MLP module, a reversible flow module and a sorting optimization module, and setting a loss function of the reference-free image quality evaluation model; inputting the training set into the image quality evaluation model, and carrying out the iterative updating of model parameters through back propagation; and inputting a test image into the trained image quality evaluation model to obtain a quali