Enhanced shadow removal using space-variant anisotropic PDEs and tensor-based osmosis models

Abstract

In this work, we propose an innovative approach to shadow removal in images by enhancing the space-variant anisotropic Partial Differential Equation (PDE) framework. This enhancement involves redefining existing osmosis models to incorporate tensor-based characteristics inspired by Weickert-type operators. The model leverages spatially distant pixel similarities, using a specified window size to improve image restoration. The numerical solution is obtained through the Split Bregman Iterative method (SBI). To thoroughly evaluate the model’s performance, we conduct a series of tests, including both perceptual and quantitative assessments. The results demonstrate that our approach outperforms state-of-the-art methods in shadow removal.