A new perceptual-based no-reference contrast metric for natural images based on human attention and image dynamic

摘要

Contrast measurement is one of the most important tasks in image quality assessment and image enhancement. However, a generic metric to evaluate contrast quality is difficult to design because different applications have different definitions of contrast. In this paper, a new perceptual-based contrast metric for natural images is proposed. Departing from traditional metrics, the proposed measure is based on both the luminance and chrominance information of images. The technical assessment of luminance contrast is mapped onto a perceptual-based measure via a human attention model. Local luminance contrast is also measured to overcome the limitations of the global measure. The CSIQ and the TID2008 databases are used for evaluating the performance of the proposed metric. It is shown that the proposed approach yields a significant improvement in comparison to the state-of-the-art.