Face image super-resolution method based on heat map prior guidance

The invention discloses a face image super-resolution method based on heat map prior guidance. According to the method, a network model structure is composed of a super-resolution trunk branch and a heat map prior estimation branch. The super-resolution trunk branch receives a low-resolution input image, and outputs the low-resolution input image after super-resolution reconstruction to obtain a high-resolution image; the prior estimation branch uses the output of the super-resolution trunk branch as the input, estimates a key point, calculates a thermodynamic diagram, and then feeds back the thermodynamic diagram as the structure prior information to the super-resolution trunk to guide the super-resolution reconstruction. In the training stage, the two branches cooperate with each other and are optimized together. In order to maximize the guiding effect of heat map prior, heat map information is utilized from two levels. And finally, a discrimination network is introduced to supervise the training of the rec