Fast Segmentation-Based Backlight Dimming

Abstract

A fast segmentation-based backlight dimming is proposed using a human perception-based three dimensional (3D) histogram. The conventional segmentation-based backlight-dimming technique requires high computation time, and cannot change the number of segments, which in turn degrades the image quality. The proposed method significantly reduces the computation time using a histogram-based segmentation approach and dynamically changes the number of segments using the peak extraction. Moreover, it uses the RGB-combined histogram for calculating the acceptable candidate error to further reduce the computation time. It can select the most suitable trade-off point between the image quality and power consumption while requiring low computation time. Based on the experimental results, the proposed method appropriately selected the trade-off point between the image quality and power consumption in images with full high-definition resolution and reduced the computation complexity by up to 43.551 s (75.73%) compared with the benchmark method.