Spectral Domain Spline Graph Filter Bank

In this paper, we present a structure for two-channel spline graph filter bank with spectral sampling (SGFBSS) on arbitrary undirected graphs. Our proposed has many desirable properties; namely, perfect reconstruction, critical in domain, flexibility choice of shape and cut-off frequency the filters, low complexity implementation synthesis section, thanks to our closed-form derivation its sparse structure. These properties play pivotal role multi-scale transforms signals. Additionally, framework can use both normalized non-normalized Laplacian any graph. We evaluate performance SGFBSS nonlinear approximation denoising applications through simulations. also compare method existing structures show superior performance.