Research Article
A Hierarchical Smoothing Method for Animation Image Based on Scale Decomposition
@INPROCEEDINGS{10.1007/978-3-031-50549-2_2, author={Jieling Jiang and Wei Li}, title={A Hierarchical Smoothing Method for Animation Image Based on Scale Decomposition}, proceedings={Advanced Hybrid Information Processing. 7th EAI International Conference, ADHIP 2023, Harbin, China, September 22-24, 2023, Proceedings, Part III}, proceedings_a={ADHIP PART 3}, year={2024}, month={3}, keywords={Scale Decomposition Animated Image Hierarchy Smoothing}, doi={10.1007/978-3-031-50549-2_2} }- Jieling Jiang
Wei Li
Year: 2024
A Hierarchical Smoothing Method for Animation Image Based on Scale Decomposition
ADHIP PART 3
Springer
DOI: 10.1007/978-3-031-50549-2_2
Abstract
In order to solve the problem of low quality of animated images affected by noise, a hierarchical smoothing method for animated images based on scale decomposition is proposed. Get the animation base image, and obtain the detail layer of the source image and target image. Use U-net convolutional neural network to select the decomposition box, select the results according to the remote sensing image segmentation box, and design the image decomposition process. Adjust the animation decomposition scale, focus on measuring multi-scale morphology, use the mean coordinate method to fuse the brightness of the target image, and retain rich details of the image. The fusion image smooth mosaic processing flow is designed, and the minimum variance standard is used to obtain the best matching combination. The gradient is used to represent the direction and size of the pixel changes in the animation image, and the details of the animation image are enhanced by means of superposition correction to achieve image edge smoothing. The experimental results show that the image details obtained by this method are consistent with the image samples, the signal-to-noise ratio is above 90 dB, and the longest smoothing processing time is 27 s, which can obtain high-quality animation images.
