Optimal liquid crystal display backlight dimming based on clustered contrast loss

We propose an image adaptive backlight dimming method that quantitatively measures the perceived image quality degradation in terms of brightness and contrast. Unlike conventional methods, the proposed adaptive dimming considers the spatial distribution characteristics of the clipped pixels via a new measure, clusterization, to effectively estimate the perceived contrast loss and prevent the clipping artifact (light saturation). The proposed adaptive dimming achieves an average 17.71% power reduction while keeping the image quality difference to a tolerably low amount, as shown by the subjective mean opinion score test results. Comparing the optimal backlight levels estimated by the proposed method with results from other methods, the proposed backlight dimming is closer to the ground truth backlight levels that are favored by human subjects.