Scene text image super-resolution method and system based on mixing

The invention discloses a scene text image super-resolution method and system based on mixing, and relates to the technical field of image reconstruction, and the method comprises the steps: transmitting a low-resolution image to a trained image network model, so as to output a super-resolution result of the image; the training of the image network model comprises the following steps: constructing a training set, and performing image reconstruction on a low-resolution scene text image to obtain a general super-resolution image; textboxes are extracted from the low-resolution scene text image and the general super-resolution image respectively, and then corresponding text line images are obtained through zooming processing; for each low-resolution text line image, generating a corresponding reconstructed super-resolution text line image; fusing each reconstructed super-resolution text line image into a super-resolution text line image position to obtain a super-resolution result of the low-resolution scene tex