Research On Liquid Crystal Display Dimming Issues Based On Image Analysis

After many years development, liquid crystal display has become main display technology used in the fields of electro-communication, household appliances, aerospace, industrial control, automotive electronics, and so on. As the fact that liquid crystal itself can’t emit light, backlight unit is necessary to provide light. As one kind of non self emitting display, liquid crystal display uses back light module as the light source. Backlight unit mianly consists of LED and a stack of optical films, which can be divided into direct-lit type and edge-lit type according to different LED setting location. The luminance of common backlight unit is always constant, which results in excessive power consumption and light leak problems. Therefore, dynamic dimming technology is proposed, which can control the luminance of backlight unit dynamically according to image content and compensate liquid crystal pixel, leading to power consumption reduction and contrast ratio improvement. Based on different control methods of backlight unit, dynamic dimming technology can be divided into global dimming and local dimming.This dissertation aims at deep analysis of current dynamic dimming technologies and tries to find core issues of nowdays dimming algorithms. Based on relevant theories of digital video processing, the dissertation tries to combine the video splitting methods and dynamic dimming technology; image characteristics extraction and image intensification are explored to improve the algorithms of backlight dimming and liquid crystal pixel compensation. The prototypes based on both global and local dimming algorithms are developed. The main research works include:1) Advantages and disadvantages of current backlight extraction methods are analyzed synthetically. A valuable evaluation system is established, espectically suitable for dynamic dimming algorithm. Relationship between image characteristics and backlight dimming level is also established via the method of statistical. A backlight dimming algorithm base on histogram and dominant gray level is therefore proposed.2) As the fact that S-curve-based dimming algorithm presents no enough adaption for various typies of image, image-type-based dimming algorithm is proposed, which classifies image effectively with deep considerations of original image typies and human display. The proposed algorithm presents the merits of better display effect and lower power consumption. With the application of a-value smoothing mechanism, the flickers can greatly improve, especially caused by rapid variation of s-type curve parameters.3) The multiple-block local dynamic dimming algorithm is proposed in order to adapt different backlight structure and block-control methods. The backlight dimming level of each block is decided by utilizing partition images from multiple-block. In regard to back light assignment problem existing in edge-emitting local dimming algorithm, quasi-optimal back light value assignment scheme is solved through simulated annealing algorithm based on minimum power consumption principle.4) A backlight dimming algorithm based on shot characteristics is proposed. The algorithm uses the brightness, main color histogram and color variance parameters to split video. The video type of every shot is confirmed, to select suitable dimming parameters according to video type.5) Based to the proposed algorithms in this dissertation, three dynamic dimming prototypes are developed. The display effect and power saving are practically measured.