Image compression recovery method and system based on multi-head heterogeneous convolution auto-encoder

The invention discloses an image compression recovery method and system based on a multi-head heterogeneous convolution auto-encoder, and the method comprises the steps: carrying out the processing of an input original image based on a heterogeneous transformation method, and obtaining a heterogeneous image; based on a deep learning method of a convolutional auto-encoder, encoding the original image and the isomerized image to obtain an original image code and an isomerized image code; fusing and quantifying the original image code and the heterogeneous image code based on an attention mechanism to obtain a compressed image; decoding the compressed image based on a deep learning method of a decoder to obtain a recovered image; and constructing a loss function based on the difference between the restored image and the original image, and continuously iterating to convergence by training the loss function to obtain an optimal restored image. According to the method, heterogeneous processing is carried out on th