Previous methods for absolute phase restoration of single frame animated images have poor restoration performance due to only grayscale processing of the obtained single frame animated images.Therefore,a method for absolute phase recovery of animation sin-gle frame images was designed that integrates self supervised learning.Animated single frame images are obtained from multiple websites,and perform grayscale processing,smoothing processing,image enhancement,and normalization on the images to construct an animated single frame image dataset.Under the action of self supervised learning,transform the animated single frame images,calculate the image activation function,extract image features under different transformation states,and obtain phase information of the animated single frame images,construct a corresponding image absolute phase recovery model,solve the model,and map the solution results to achieve absolute phase recovery of a single animation frame image.In experimental testing,compared with previous methods for absolute phase recovery of animation single frame images,the designed animation single frame image absolute phase recovery method that integrates self supervised learning has a higher peak signal-to-noise ratio and better recovery effect in practical applications.
fusion self supervised learninganimated single frame imageabsolute phaseimage restorationrecovery methods
维普
万方数据
