Gaussian Fourier Pyramid for Local Laplacian Filter

Multi-scale processing is essential in image and computer graphics. Halos are a central issue multi-scale processing. Several edge-preserving decompositions resolve halos, e.g., local Laplacian filtering (LLF), by extending the pyramid to have an property. Its costly; thus, approximated acceleration of fast LLF was proposed linearly interpolate multiple pyramids. This paper further improves accuracy Fourier series expansion, named LLF. Our results showed that has higher for same number Moreover, exhibits parameter-adaptive property content-adaptive filtering.