抄録
This paper extends the existing TV-G-L^2 nonlinear image decomposition approach to nonlinear color-image decomposition. For utilizing inter-channel color cross-correlations, this paper first introduces the TV norms of color differences and color sums into the TV-G-L^2 functional, and then derives an optimization algorithm using the wavelet transform. Our proposed method decomposes color images without producing color artifacts, and achieves high-quality decomposition robust against random noise.
