Unsupervised enhancement method for low-quality face image and storage medium

The invention discloses an unsupervised enhancement method for a low-quality face image and a storage medium, and belongs to the technical field of image enhancement. Brightness enhancement of a non-uniform face image is realized by constructing an enhancement model and analyzing face image information and an illumination state, and the enhancement model is based on deep learning and neural network technologies. The method can automatically learn the features of a face image and an enhancement method, improves the quality of the image, constructs a discrimination model by using a generative adversarial network model, processes the output of an enhancement model in combination with a constraint function so as to maintain the real details and natural image state of an enhancement result, and carries out the effective constraint of the enhancement result. According to the method, the face feature information in the image is fully mined, the face state of natural exposure is kept, the face information of the low-