High-visibility tone mapping by weighted guided image filter in multi-scale domain

In this paper, we propose a novel tone mapping method to obtain high visibility by boosting local luminance contrast with guided image filter. Specifically, the original images are decomposed into multi-scale layers by the weighted least square filter. A novel pixel-wise weighting map employing the guided image filter on each layer is presented. The key idea is to assign large weightings to the regions with large contrast such that lighting effects are highly reduced and halo artifacts merely occur. Moreover, color balancing correction is performed to suppress color distortion. Experiments on several testing sets demonstrate the proposed method preserves clearer details and more faithful color tone than the existing methods in terms of visual and objective evaluation.