Universal image restoration method based on automatic screening mechanism

• Main Authors: CHENG SHICHAO, WANG YIFAN, ZHOU JUNZHE, ZHANG JIANHAI
• Format: Patent
• Language: Chinese; English
• Published: 10.05.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

The invention discloses a general image restoration method based on an automatic screening mechanism, and the method comprises the steps: firstly downloading images from a public data set, randomly selecting n images with different damage degrees to form an image set, dividing the image set into a training set and a test set according to a proportion, and carrying out the preprocessing of the images in the training set and the test set; secondly, extracting features from the preprocessed training set data by using a residual network, and judging a to-be-repaired type of the training set data; and fine tuning is carried out on different pre-training parameters in the multi-task image restoration model to realize degraded image restoration. And finally, calculating loss to obtain optimal parameters of the residual network and the multi-task image restoration model, and testing. According to the invention, a high-efficiency and high-quality image restoration function is realized, the use cost of an image restora