Deep Image Prior with L0 Gradient Regularizer for Image Smoothing
This addresses the problem of image smoothing for computer vision applications by eliminating the need for training data, though it is incremental as it builds on deep image prior methods.
The paper tackles the challenge of image smoothing without requiring curated training data by proposing DIP-ℓ0, a deep image prior framework with an ℓ0 gradient regularizer, which outperforms existing algorithms in edge-preserving smoothing and JPEG artifact removal.
Image smoothing is a fundamental image processing operation that preserves the underlying structure, such as strong edges and contours, and removes minor details and textures in an image. Many image smoothing algorithms rely on computing local window statistics or solving an optimization problem. Recent state-of-the-art methods leverage deep learning, but they require a carefully curated training dataset. Because constructing a proper training dataset for image smoothing is challenging, we propose DIP-$\ell_0$, a deep image prior framework that incorporates the $\ell_0$ gradient regularizer. This framework can perform high-quality image smoothing without any training data. To properly minimize the associated loss function that has the nonconvex, nonsmooth $\ell_0$ ``norm", we develop an alternating direction method of multipliers algorithm that utilizes an off-the-shelf $\ell_0$ gradient minimization solver. Numerical experiments demonstrate that the proposed DIP-$\ell_0$ outperforms many image smoothing algorithms in edge-preserving image smoothing and JPEG artifact removal.