Constrained Adaptive Nonlinear Unsharp Masking

Ciprian Orhei, Ismet Porobic, Adrian Avram

Abstract: Image enhancement is one of the most important topics in the domain of computer vision, particularly due to the expansion of mobile digital imaging. Digital image quality often suffers from factors such as inadequate contrast, undesirable shading, artifacts, or poor focus. To counteract these issues, significant research has been invested in image de-noising and sharpening to enhance the quality of captured images. In this paper, we present the Constrained Adaptive Nonlinear Unsharp Masking, a novel approach to the classical Unsharp Masking that involves controlled adaptive and nonlinear adjustments to the enhancement features across the image. The proposed algorithm has been evaluated with positive results on both synthetic and natural images.