In this letter, a new backlight dimming technique that considers perceptual image quality based on a structural similarity (SSIM) metric is proposed. Conventional algorithms cannot accurately control the optimal trade-off between image quality and power consumption; image quality is therefore degraded in some images. In contrast, the proposed algorithm calculates the allowable target error based on the SSIM metric in real time using the relation function between SSIM and the mean square error (MSE). It can therefore optimally reduce power consumption while preserving the perceptual image quality over the target level. Experimental results show that the proposed algorithm successfully maintains the target SSIM for all test images and sequences. Additionally, the average SSIM of the proposed algorithm is up to 0.034 higher than those of the benchmark algorithms.
© 2014 IEEEPDF Article