Image-classification-based Global Dimming Algorithm For LED Backlights In LCDs

As the fact that liquid crystal devices can’t emit lights, backlight units are necessary.Backlight units usually work under the condition of presenting max luminance, whichmay cause power waste and lower contrast ratio. Although some energy saving rate canbe obtained by some existing global dynamic dimming algorithm, most of which alsolead to pixels overflow distortion and low contrast. S-curve-based global dimmingalgorithm proposed by predecessor can not only reduce energy consumption but alsoimprove contrast with no pixel overflow distortion. With the same image pixelcompensation parameters and backlight dimming coefficients, it is not possible for theprevious algorithms to achieve the balance between power saving and contrastimprovement.In this paper, global dimming algorithm based on image classification is proposed,which classifies images by brightness and contrast. Different type images exploredifferent pixel compensation parameter and backlight formula coefficient. On the basisof the traditional image Structural Similarity Index Measurement (SSIM) evaluationmethod, a new image quality assessment method specifically applied for the dynamicdimming algorithm in the paper was developed. With the simulation analysis, thecontrast, energy-saving and image quality evaluation value of the dimmed image werereached. And the prototype was developed and experimental test was carried out. Boththe simulation results and practical test results show that the algorithm can save energyand improve contrast effectively.